The impact of public services on health care and morbidity: evidence from Kenya

The paper reports findings made by Appleton et al. (1991) in a World Bank research project `Women and Public Services in Africa', for which funding is gratefully acknowledged. However, it represents the views of the author alone and should not be attributed to the World Bank. The author is employed by the Centre for the Study of African Economies, a Designated Research Centre of the Economic and Social Research Centre. The author would like to thank Steve Bond for advice on this topic and Guiseppe Mazzarino for writing the NAG program for numerical evaluation of double integrals. The work extends and refines the analysis in Bevan, Collier and Gunning (1989); the authors of that book provided much useful discussion as well as access to their data. Helpful comments were received from reviewers of the World Bank project and participants at seminars in Oxford and Washington DC. Health care and status are jointly modelled using household data from Kenya. Both maternal primary education and distance to health facilities affect take-up of child health care, but the former is more powerful. Corrections for selectivity when modelling health demand of the sick are insignificant. Parental education, distance to health facilities and piped water all increase, reporting of illness symptoms. The first two results are interpreted as reporting biases but the effect of piped water remains disturbing. The impact of health care on the duration of illness is estimated controlling for endogeneity and found to be favourable but insignificant.