Achieving child-health-related millennium development goals: The role of infrastructure - A comment

Using the differences observed across countries and asset quintiles within countries, Fay, Leipziger, Wodon, and Yepes argue that basic infrastructure brings large reductions in child mortality and stunting. Their estimation method makes a number of questionable assumptions, some of which can be readily relaxed with the same data. On using an alternative estimator and augmenting their data set to include female schooling, one finds little sign that better infrastructure lowers either child mortality or stunting. The infrastructure impacts found by Fay et al. stem from a combination of functional-form misspecification, latent country effects, and omitted quintile-specific schooling effects. (c) 2007 Elsevier Ltd. All rights reserved.