The international diversification puzzle is not as bad as you think

In the data country portfolios are heavily biased toward domestic assets. Standard one-good international macro models predict that, due to the presence of non-diversifiable labor income risk, country portfolios should be heavily biased toward foreign assets; this discrepancy constitute the international diversification puzzle. (Baxter and Jermann, 1997). We show that a simple extension of one-good models help reconcile theory and data. In particular we analytically solve for the equilibrium country portfolios in a two-country, two-goods model with non-diversifiable labor income and investment. In this set-up, consistently with the data, country portfolios contain a relatively small, but positive, share of foreign assets. The reason why international diversification is low is that terms of trade movements provide considerable insurance against country specific shocks and labor income risk (Cole and Obstfeld 1991, Acemoglu and Ventura, 2001). The reason why international diversification is positive is that foreign assets are crucial to share the financing of investment across countries. Finally in the model a country\92s share of foreign assets should depend on its import share and on its capital share. We show how this relation is qualitatively and quantitatively consistent with country portfolios in the cross section of OECD countries in the 1990s.