| Employing closed-economy models, recent cross-country growth literature seems to have confirmed the conditional convergence hypothesis (CCH): holding population growth and capital accumulation constant, poor countries tend to grow faster than rich countries. However, this literature reveals an empirical puzzle. African and Latin American counties grew systematically slower than the sample mean during the 1970s and 1980s. This thesis reexamines the CCH under the assumption of open economies with imperfect capital mobility. Two alternative models are constructed. The first shows that the CCH can be extended to open economies if foreign borrowing can be used only to finance the accumulation of physical (but not human) capital. Furthermore, external variables such as debt and openness are expected to affect growth, either directly, or indirectly by affecting investment share. The second model studies the growth of a small borrowing economy facing a credit ceiling internationally. This framework classifies a borrowing economy as one of three cases: never-constrained, ever-constrained and optimal-regime-switching. The key result is that growth paths of output, investment, and foreign debt for a regime-switching country exhibit kinks and different convergence properties. This attributes low growth rates in Africa and Latin America to excessive borrowing in the late 1970s and subsequent regime-switching. Empirically, for 98 countries from 1960 to 1986, we find that openness has a positive effect on the growth rate while debt has a negative effect, and that CCH does indeed hold in an open-economy setting. Secondly, when growth and investment are both treated as dependent variables and reestimated in a simultaneous equation system, we find a significant negative effect of debt on investment share, and a much larger coefficient of investment on growth. We also reconfirm the two-link chain previously identified in the literature: openness is positively correlated with the investment share and is therefore growth-promoting. To allow for dynamic variations across time, we further respecify the model in a pooled cross-sectional and time-series analysis, producing more efficient parameter estimates. Finally, we propose a Logit model to determine the probability of a country's being credit constrained. We find that debt-service ratio, reserves-to-import ratio, average interest on the loans, lagged growth rate, lagged investment share, and share of government spending are important indicators of debt servicing capacity. |