We examine deviations from trend of net exports and other components of GNP for the United States and attempt to build models consistent with their behavior. The most striking fact is that net exports have consistently been countercyclical. We show, first, that dynamic pure-exchange models can only produce a negative correlation between net exports and GNP if the variance of consumption exceeds that of output. In (he United Slates it does not, so this class of models cannot explain observed comovements between output and trade. We then examine government spending and nontraded goods as potential remedies, but show that their behavior is either inconsistent with the data or can be made consistent with any pattern of comovements. The most promising model introduces production and capital formation. Fluctuations are driven by country-specific productivity shocks, in which high productivity domestically leads to high domestic investment and a deficit in the balance of trade. This theory also receives support from the large negative covariance between net exports and investment in American data.

We derive the empirical implications of a popular class of international macroeconomic models. The real economy is a stochastic exchange model with complete markets. A standard result is that cross-country risk sharing implies perfect correlation between consumption paths across countries. With mild restrictions on the endowment process ii also implies a positive correlation between net exports and output in every country. We introduce money using cash-in-advance constraints and show that the implications for real variables carry over into the monetary economy. These dichotomy and neutrality propositions generalize those in the literature to stochastic environments with heterogeneous agents, and do not require the cash-in-advance constraint to bind in every state. They imply that any correlation between the nominal exchange rate and the balance of trade can be made consistent with the theory.

We derive the empirical implications of a popular class of international macroeconomic models. The real economy is a stochastic exchange model with complete markets. A standard result is that cross-country risk sharing implies perfect correlation between consumption paths across countries. With mild restrictions on the endowment process ii also implies a positive correlation between net exports and output in every country. We introduce money using cash-in-advance constraints and show that the implications for real variables carry over into the monetary economy. These dichotomy and neutrality propositions generalize those in the literature to stochastic environments with heterogeneous agents, and do not require the cash-in-advance constraint to bind in every state. They imply that any correlation between the nominal exchange rate and the balance of trade can be made consistent with the theory.

We examine deviations from trend of net exports and other components of GNP for the United States and attempt to build models consistent with their behavior. The most striking fact is that net exports have consistently been countercyclical. We show, first, that dynamic pure-exchange models can only produce a negative correlation between net exports and GNP if the variance of consumption exceeds that of output. In (he United Slates it does not, so this class of models cannot explain observed comovements between output and trade. We then examine government spending and nontraded goods as potential remedies, but show that their behavior is either inconsistent with the data or can be made consistent with any pattern of comovements. The most promising model introduces production and capital formation. Fluctuations are driven by country-specific productivity shocks, in which high productivity domestically leads to high domestic investment and a deficit in the balance of trade. This theory also receives support from the large negative covariance between net exports and investment in American data.

Previous literature has shown that the study and characterization of constrained efficient allocations in economies with limited enforcement is useful to understand the limited risk sharing observed in many contexts, in particular between sovereign countries. In this paper we show that these constrained efficient allocations arise as equilibria in an economy in which private agents behave competitively, taking as given a set of taxes. We then show that these taxes, which end up limiting risk sharing, arise as an equilibrium of a dynamic game between governments. Our decentralization is different from the existing ones proposed in the literature. We find it intuitively appealing and we think it goes farther than the existing literature in endogenizing the primitive forces that lead to a lack of risk sharing in equilibrium.