Who Gains More from Which Infrastructure in Rural People’s Republic of China?
This paper will demonstrate the deficiency of conventional approaches to modelling inequality; extend the Mincer earnings function so that both growth and distributive effects of infrastructure can be evaluated.
The importance of infrastructure in economic development has been increasingly recognized by governments, development institutions, and the research community. Despite a sizable literature on its efficiency and growth effects, the distributive impacts of infrastructure have been largely overlooked, with a few recent exceptions. This is regrettable, particularly given the overwhelming concern about inequality and inclusive growth all over the world. This paper will: (i) demonstrate the deficiency of conventional approaches to modelling inequality; (ii) extend the Mincer earnings function so that both growth and distributive effects of infrastructure can be evaluated; and (iii) fit the extended model to a large sample of individual-level data from rural People’s Republic of China (PRC) over the period of 1989–2011, providing estimates of growth and the distributive impacts of specific physical infrastructures—telephone, tap water and electricity. All these infrastructures are found to promote rural income growth, helping narrow the rural–urban gap, which is the dominant component of the PRC’s overall inequality. Further, the poor are found to gain more than the rich, implying benign distributive effects of these infrastructures. In addition, males, the more experienced, the better educated, and to some extent the married benefited more than their counterparts, especially from telephones. Finally, some of these subpopulation effects have become more significant in recent years and are larger in central PRC, possibly because infrastructure helps open up more opportunities for those with better education or more experience. The empirical results are robust to different definitions of the experience variable, consideration of the mortality selection bias, reconstruction of the telephone data, and possible reverse causality.