Paradoxically, high-growth, high-investment developing countries tend to experience capital outflows. This paper shows that this allocation puzzle can be explained simply by introducing uninsurable idiosyncratic investment risk in the neoclassical growth model with international trade in bonds, and by taking into account not only TFP catch-up, but also the capital wedge, that is, the distortions on the return to capital. The model fits the two following facts, documented on a sample of 67 countries between 1980 and 2003: (i) TFP growth is positively correlated with capital outflows in a sample including creditor countries; (ii) the long-run level of capital per efficient unit of labor is positively correlated with capital outflows. Consistently, we show that the capital flows predicted by the model are positively correlated with the actual ones in this sample once the capital wedge is accounted for. The fact that Asia dominates global imbalances can be explained by its relatively low capital wedge.