Title: Poverty, GDP per capita and health in developing countries: Panel cointegration results with selection
Authors: Christophe Muller - Aix-Marseille School of Economics (France) [presenting]
Cyrine Hannafi - University Paris I (France)
Abstract: Cointegration relationships between poverty, economic growth and health are estimated by using country-level data from developing countries, using Depth of Food Deficit, GDP per capita and Infant Mortality indicators as proxies. As opposed to the current literature, we account for non-stationarity issues, distinguish between short and long term interactions, and select countries for which all the indicators are non-stationary. The determinants for selection are explored with Probit models that show that, although the occurrence of stationarity is dominated by randomness, some shock variables (on food availability, measles and natural disasters) retain some prediction power of stationarity. We conduct a panel data cointegration analysis over the selected subsample. The VECM results are compared with those without sample selection and with those obtained for time series analyses for each country, and impulse response functions are examined. The cointegration results are found to vary with the estimation approach, and affected by sample selection. For a limited proportion of countries, the results exhibit strong serial correlations of poverty and infant mortality, and a substantial impact of the error correction term on the dynamics of all the studied variables. Finally, these statistical results are confronted with an economic theory model that provides an interpretation grid.