| In chapter 2, we assess pairwise income per capita convergence for 15 OECD countries using a time series based test. Our model allows income per capita differentials to exhibit long range dependence, which generalizes the specification in previous studies. Using semi-parametric estimation procedures we find ample evidence of pairwise convergence among 15 OECD countries. This result is contrary to the literature that uses unit roots and cointegration tests to analyze income convergence. In chapter 3, we propose a novel approach to estimate and make inference from growth equations. We use quantile regression to assess income convergence and the effects of policy variables on the conditional distribution of the GDP growth rates. Our findings suggest that previous empirical growth studies relying on conditional mean estimation methods such as least squares give a misleading picture of the growth dynamics. Furthermore, it is suggested that the underlying growth model generating the growth experience of fast-growing countries is of neoclassical type while for slow-growing counties is of endogenous growth type. In chapter 4, we use a continuous-time stochastic overlapping generations model to explain cross-section variability in real interest rates. We find that agents with a higher life expectancy accept a lower return on the risk-free asset. Moreover, agents with relatively large share of wealth invested in human capital will hedge against human capital (idiosyncratic) risk, and therefore, accept a lower return on the risk-free asset. Empirical evidence for OECD countries supports the main implications of the model. Finally, we provide estimates of the agents' degree of risk aversion, elasticity of intertemporal substitution and subjective rate of time preference. |
