Essays on the Determinants of Capital Flows

This thesis investigates the determinants of international capital flows and strives to present new evidence-based answers to the long-standing question of why capital tends not to flow from rich to poor countries as predicted by standard neoclassical theory – a puzzle known as the Lucas paradox. This thesis consists of four stand-alone empirical studies, each of which builds an inherently coherent story exploring a possible answer to the Lucas paradox motivated by the goal of empirically identifying the determinants of international capital flows. Chapter 2 provides contextualization of the following four empirical chapters, first reviewing the predictions of standard neoclassical theory and then the vast empirical literature describing and attempting to resolve the Lucas paradox. The first of the four empirical studies appears in Chapter 3; it reviews alternatives to conventional estimates of the marginal product of capital (MPK) and examines whether these revised estimates of MPK succeed in helping us understand observed international capital flows. Bivariate data analysis (i.e., using scatter plots of capital inflows and revised MPK estimates) suggest that the direction of capital flows points, if anything, towards relatively low-return countries, a pattern that runs opposite to the prediction of standard theory. Instead of solving the Lucas paradox, these revised MPK estimates recast the puzzle from Lucas’ original question of ‘Why doesn’t capital flow to poor countries?’ to a new one: Why does capital flow to low-return countries? One possibility is that the revised MPK estimates do not capture adequately all the important factors that influence international capital flows, implying a need for further analysis that motivates this thesis. The second empirical study in Chapter 4 looks for a solution to the Lucas paradox using cross-sectional regression analysis with new specifications of the conditional mean capital inflow function. By replicating a closely related empirical work (Alfaro, Kalemli-Ozcan, and Volosovych, 2008), we show that differences in institutional quality (e.g., property rights) between rich and poor countries are not sufficient to explain fully why such large capital flows are observed flowing from poor to rich countries. We argue that, when the empirical model is specified more appropriately, differences in institutional quality cannot fully explain why capital flows to rich countries. We confirm this finding using an updated dataset on capital inflows covering the most recent decade. This updated analysis shows that institutional quality alone does not solve the Lucas paradox because capital is still observed flowing to rich countries. The third empirical study presented in Chapter 5 examines whether international capital flows can be explained using panel data and panel model estimators to filter out effects of more complex intertemporal dependencies while controlling for country-specific omitted variables and addressing the potential endogeneity of these determinants. We reach the same conclusion that differences in institutional quality between rich and poor countries cannot fully explain why capital does not flow from rich to poor countries. Moreover, panel-data estimates suggest that generous policies relating to capital account convertibility (i.e., lower levels of restrictions on capital movements) are relatively more influential than differences in institutional quality in determining international capital flows. In Chapter 5, we also examine the claims in the literature that interaction effects (i.e., multiplicative effects of two independent determinants) are required to explain or account for the Lucas paradox. Having analysed several specifications allowing such interaction effects, our findings on the explanation of the Lucas paradox remain unchanged. Capital tends to flow to rich countries and economic development (measured by GDP per capita) is a primary determinant of capital inflows. Institutional quality, capital account openness, the stock of human capital, and/or other potential factors examined also appear as significant determinants, although the size and statistical significance of estimated marginal effects depend on the estimation approach. These factors do not, however, fully capture the effect of economic development. The Lucas paradox therefore stands, as yet, unresolved. In the fourth and final empirical study presented in Chapter 6, we examine the relative importance of various political risks (e.g., expropriation risk, corruption, lack of democracy etc.) on foreign direct investment (FDI) flows into developing countries. We evaluate the percentage change in expected FDI inflows associated with improvements (e.g., one-standard-deviation and worst-to-best improvements) in various political risk measures and find that expropriation risk (i.e., the probability of losing 100 percent of invested capital) has among the largest effects on FDI, suggesting an intuitively appealing explanation for the lack of FDI flows to many capital-scarce developing countries