Health, safe water and sanitation: a cross-sectional health production function for central Java, Indonesia

Abstract

The study describes the development of health production functions and their application in the evaluation of the health impacts of investments in safe water and sanitation. For this purpose, data on the morbidity of waterborne diseases and diarrhoea were collected from medical records in the province of Central Java, Indonesia. A reciprocal production function was found to fit the data best. The health production functions exhibit constant return to scale, i.e., a simultaneous m-fold increase in both safe water and sanitation coverage produces a 1-1/m decrease in morbidity. Safe water was found to be more important for health than the sanitary disposal of excreta.

PIP: A review of prior studies has shown that safe water supply and sanitation (WSS) improvements can decrease diarrhea morbidity and mortality, but most of the studies are flawed by methodological deficiencies, inadequate health indicators, and failure to control for confounding factors. In this study, a production function approach at the community level is used to determine the effects of WSS. Morbidity from diarrhea and all water-borne diseases is the dependent variable. Many other variables which impact on morbidity were not included in the analysis due to unavailability. The production function was specified as follows: morbidity of water-born diseases from January to December 1990 in Indonesia was equal to the function of safe water supply coverage and access to sanitation. Diarrhea morbidity is separately specified. The data was fitted to linear, quadratic, reciprocal, log-linear, reciprocal log-linear, and double-log (Cobb-Douglas) production functions. Ordinary least squares (OLSQ), and generalized least squares (GLS) methods, when heteroscedasticity was present, were used. Specifications were compared in a variety of ways; e.g. generalized cross validation. Data were obtained for 14 districts and municipalities (194 subdistricts) in Central Java, Indonesia. Mean safe water coverage was 56% and access to sanitation was 61% of the population. Equations with the constant term were insignificant for WSS. Only the reciprocal functions of OLSQ without a constant term showed significance for both water and sanitation. GLS methods improved regression results. Isoquant maps to show morbidity were drawn. The conclusions were 1) that a minimum of safe water or sanitation is necessary in order to achieve a given level of morbidity; 2) there is limit beyond which morbidity conditions cannot be improved with WSS; 3) the existing mean coverage of WSS of 56% and sanitation of 39% is well below the minimum needed to minimize morbidity, if there is no other controlled variables. Safe water is more important than sanitation, but there is low substitutability. Both must be improved simultaneously. Estimates are provided of the reduction in morbidity associated with specific increases in WSS. Even 100% coverage will not eradicate water-borne or diarrhea diseases.