Duke 2008 | ASHE | Equity and Efficiency in Health and Healthcare

While it is well established that non-white infants are more likely than white infants to die within the first year of life, we know little about the complex set of factors that are driving this difference. This paper is an empirical investigation of the racial and ethnic differences in birthweight, an important outcome to study since low birthweight is the perhaps the single most significant predictor of mortality within the first year of life. Prenatal care use, mother's socioeconomic status and the availability of health services are important inputs into the birthweight production function; and each of these inputs have been found to explain, to some extent, the persistent gap in white and non-white birthweight. Although each of these inputs has been studied individually, what remains unanswered in the literature so far is to what extent do white/non-white differences in observable characteristics such as health behaviors, maternal demographics, and the supply of health care in the local community, explain the difference in white and non-white birthweights in the U.S.?

In this paper we apply the Oaxaca-Blinder decomposition method, a tool generally used in the study of labor market discrimination, to decompose the racial and ethnic differences in birthweight into the component that is explained by differences in observable characteristics and the component that is due to other factors. The observable component is divided into three categories: mothers' prenatal care decisions, parental demographic and socioeconomic status, and the availability of health services. Data are obtained from several sources in order to capture as many observables as possible. We thus estimate the magnitude of the racial difference in birthweight that is attributable to differences in the inputs for non-white babies as compared to white babies. The residual or unexplained difference provides an estimate of the extent to which other factors, such as racial differences in productivity of inputs into the health production function and discrimination, may be driving white/non-white gaps in birthweight and, by extension, infant mortality.

The 2000 Natality Detail File is a compilation of birth certificates of the universe of children born in the U.S. in 2000, and is the primary source of data for this analysis. In addition to parents' demographics and health outcomes at birth, the file contains information on mothers' prenatal decisions such as timing, frequency and adequacy of prenatal care use, weight gain, and prenatal alcohol and tobacco use. County-level information on the availability of health services are obtained from the 2005 Area Resource File. We conduct separate analyses comparing birthweight for black and white infants and for Hispanic and white (non-Hispanic) infants. Results suggest that whereas only 15% of the mean birthweight difference between black and white infants is explained by differences in observable characteristics, the same set of characteristics explain approximately 54% of the difference in Hispanic and white non-Hispanic birthweight. The sensitivity of these results to alternate variable definitions is addressed in the paper.