International Capital Flows with Incomplete Markets

National Bureau Of Economic Research Inc, Cambridge MA

Investigators

Abstract

The last two decades have witnessed a remarkable increase in the degree of financial integration among countries. There has been a sharp increase in both net and gross capital flows. Gross external assets and liabilities have tripled among industrialized countries as a fraction of GDP, while large net external imbalances as faced by the United States have lead to a significant debate about the external adjustment to such imbalances. Nonetheless, two and a half decades after the first models of international capital flows were developed in an inter-temporal framework, our understanding of capital flows remains very limited. Most models assume that only one risk-free bond is traded. Such a framework cannot be used to analyze gross capital flows. Moreover, changes in expected asset returns or the risk-characteristics of assets play no role since there is no portfolio choice. The goal of this project is to develop and apply a method for solving dynamic stochastic general equilibrium (DSGE) models with portfolio choice. The aim is to achieve a richer understanding of the driving forces behind capital flows and the accompanying adjustment of exchange rates, asset prices and goods prices. The methodology extends standard first and second-order solution methods for DSGE models in a way to accommodate portfolio choice. For example, it is common in models without portfolio choice to first solve the deterministic steady state, around which model equations are expanded. But portfolio choice is not well defined in a deterministic environment. It depends on second moments, such as variances and covariances of asset returns, so that even to compute the zero-order or steady state of portfolio allocation requires using the second-order component of optimality conditions for portfolio choice. Similarly, the first-order time variation of portfolio choice depends on the third-order component of the optimality conditions for portfolio choice, which captures time-varying second moments of asset returns as well as time variation in expected return differences across assets. This method has applications that reach beyond just portfolio choice, giving it broader impacts. For example, in models with precautionary savings the steady-state wealth distribution depends on the second-order component of optimality conditions for inter-temporal consumption choice which capture precautionary savings. The method is applied to two-country DSGE models, starting from a simple framework that focuses on portfolio choice alone to increasingly richer setups that introduce consumption, investment, a government and richer asset market structures. The method can also be used to fold noisy rational equilibrium models that are commonly adopted in finance, and which are generally static, into dynamic open economy setups. These models allow further richness in the form of information asymmetries across investors within and between countries. The models have implications for both gross and net capital flows and the driving forces behind capital flows. For example, under what circumstances do capital inflows and outflows move in the same direction or in opposite directions? Are capital flows mostly driven by a reallocation of portfolios across assets or by savings that is allocated across assets based on steady state portfolios? The answers to such questions in portfolio choice models can be connected to capital flows data for industrialized countries in order to learn what features of modeling are most critical to understand what drives capital flows in the data.