A Dynamic Equilibrium Model of International Portfolio Holdings: Comment By

ONE OF THE MAIN PUZZLES in international finance is portfolio home bias: typical investors hold most of their wealth in domestic assets (despite the fact that international diversification permits a sizable reduction in risk). Serrat (2001) claims that the nontradable nature of many consumption goods explains home bias. He presents a model of a world with two countries that receive tradable and nontradable endowments; financial transactions are restricted to trade in stocks that are claims to these endowments.2 Serrat solves for a Pareto efficient equilibrium of that economy. Serrat makes the following assertions: (i) The equilibrium portfolio is unique (p. 1472; Theorem 2). (ii) Claims to a country’s traded good endowment are mainly held by local investors (for plausible correlations between endowments; see Table II). (iii) Claims to a country’s nontraded good endowment are only held by local investors (Theorem 2). This note shows that those assertions are incorrect: (i) Serrat’s model fails to explain home bias. Due to the assumed preferences, dividends (evaluated at equilibrium goods prices) and equity prices are collinear in Serrat’s economy. Thus, portfolios are indeterminate: the model does not pin down what fractions of the claims to a country’s traded and nontraded good endowments are owned by local investors, what fraction of her wealth an investor allocates to domestic assets, or what fraction she allocates to claims to traded goods—these fractions can take any values. However, the model predicts that each investor’s holdings of claims to traded goods are fully diversified internationally (each investor holds a share in the domestic traded goods sector that equals her share in the foreign traded goods sector); this prediction is counterfactual. (ii) The portfolio holdings described by Serrat (Theorem 2) do not implement the efficient allocation. I next summarize the model features that are needed to establish these points.