In this paper we test the well-known hypothesis of Obstfeld and Rogoff (2000) that trade costs are the key to explaining the so-called Feldstein-Horioka puzzle. Using a gravity framework in an intertemporal context, we provide strong support for the hypothesis and we reconcile our results with the so-called home bias puzzle. Interestingly, this requires a fundamental revision of Obstfeld and Rogoff?s argument. A further novelty of our work is in tying bilateral trade behavior to desired aggregate trade balances and desired intertemporal trade.