Health plays an important role in the economic growth process and increasing productivity concurrently creates better health outcomes. The stock of an individual’s health increases as income rises as well as investment in health augmenting consumption. The increasing attention paid to the health-income nexus motivates this thesis to use macroeconomic theory, tools and methods to examine the link. This study first replicates a highly cited empirical work by Hartwig (2008) on Baumol’s Cost Disease (BCD) to explain the rising trends of healthcare expenditure. The gaps identified during the replication prompt the designing of testable theoretical models of BCD in Chapter 3. The formulated BCD hypotheses are tested using two sets of panel data estimators that assume slope homogeneity and heterogeneity. In Chapter 4, a Monte Carlo simulation experiment is conducted to investigate the performance of recently developed “mean group type” panel estimators that are robust to cross-sectional dependence, slope heterogeneity, non-stationarity and endogeneity. The simulation experiment results inform the choice of the panel estimators used to examine the income elasticity of healthcare expenditure for selected African countries in Chapter 5. Overall, I find that Hartwig’s (2008) study suffers from methodological flaws and the tested hypotheses revealno significant relationship to support the BCD predictions using an OECD dataset. From the simulation experiments, I find the Common Correlated Effect Mean Group (CCEMG) estimator to be the “best” on the dimensions of bias, efficiency, and coverage rates. Lastly, the empirical analysis using the dataset for 47 African countries and estimating with CCEMG provides no significant outcomes to support whether total, public and private health expenditures are either income inelastic or elastic i.e., health care is a necessity or luxury good.

Key words: Health expenditure, income, panel data, cross-sectional dependence, slope heterogeneity, factor models, endogeneity, OECD, Africa