The Feldstein-Horioka puzzle revisited: A Monte Carlo study

This study revisits the Feldstein-Horioka (FH) puzzle by employing a variety of asymptotically efficient cointegration estimators, and by using the critical values obtained from Monte Carlo simulations to test the hypothesis of a unit retention coefficient. This leads to a lower rejection of the null than standard critical values would have implied. Despite ample evidence supporting the FH result, there appears to be considerable heterogeneity in terms of the savings-investment association, and only 25% of the 23 OECD countries we examine can be characterised as open economies in the FH sense. Whether or not one subscribes to the FH interpretation, the FH result does not appear to be robust.