The Feldstein-Horioka puzzle revisited

Krol (1996) argues that the Feldstein-Horioka puzzle is the result of the estimation technique that using fixed-effect panel regression, low saving -retention coefficients are found. However, Krol's study has been criticized as revolving around the inclusion of Luxembourg. This paper reexamines the Luxembourg problem by applying Kao and Chiang's (2001) DOLS and FMOLS estimators to non-stationary panel data of 20 OECD countries. Our empirical evidences show that the inclusion or exclusion of Luxembourg does not affect the estimation results; instead, the power of the estimation technique and of the hypothesis test matters. This result is interpreted as the higher power of cointegration analysis in panel data.