Three essays on investment, saving, and the current account

The purpose of this dissertation is to answer two major questions. First, can we provide additional evidence pertaining the validity and usefulness of the intertemporal model of the current account? And second, if we say yes to the first question, how will this model perform in the empirical studies related to macroeconomic factors, such as investment and saving, that are closely related to the current account?;The first essay tests the intertemporal model of the current account in a large country framework. Compared to the standard small country model commonly used in the literature, our model shows that only country-specific component of net income will affect the current account, and it generates smoother current account series, implying a stronger connection between current consumption and current net income. Subsequent comparative empirical studies of small- and large-country models, done in the framework of a present value model, show that the large country framework out-performs the traditional small country framework when the object of interest is indeed a large country.;The second essay targets primarily the relationship between degree of persistence of terms of trade (TOT) shock and the current account. Our major contribution is to study this topic in a decomposition framework. To get a measure of the degree of persistence, we use two decomposition techniques, Beveridge-Nelson decomposition and HP filter, to decompose TOT into permanent and transitory components. Empirical results show that the persistence of TOT shocks does play an important role with respect to how the current account respond to such shocks.;The third essay investigates the relationship between investment and saving, or the Feldstein-Horioka (FH) puzzle, by applying regular panel estimation models to time series of investment and saving that are decomposed using a similar approach as in the second essay. Our empirical results confirm the findings of the earlier work done using different econometric frameworks that short-run correlation is generally much smaller than long-run correlation. Dependent on particular choice of decomposition technique and panel model, we find some evidence that capital is very mobile for OECD countries 1960-2004.