Effects of antenatal care services on birthweight: importance of model specification and empirical procedure used in estimating the marginal productivity of health inputs.

This paper is an attempt to examine the effects of antenatal care (ANC) utilization on birthweight. The analysis is based on the data collected from ever-married women by Turkey Demographic and Health Survey 1998 (TDHS). Cost-effectiveness based health sector reform requires careful estimation of costs and productivity of health interventions as well as the substitution possibility among health inputs. If the parameters of production function are known, policy makers can estimate the health outcome effects of various input-mix. Two-stage least squares (TSLS) and ordinary least squares (OLS) estimation procedures are used in this paper to estimate the effect of health care input, namely antenatal care visits, on birthweight. Since the utilization of medical care itself may be dependent on women's expectation about the pregnancy outcome, the parameters estimated through OLS may underestimate the true productivity of the input. The estimated functions indicate that antenatal care, woman's health status, and birth order are significant determinants in birthweight. The TSLS estimate of marginal productivity of ANC visits was about four times the marginal productivity estimate in the OLS model. The sensitivity of the parameter estimates to model specifications and empirical procedures followed demonstrates the importance of selecting the right model from both theoretical and empirical point of view.