Brookings-Wharton Papers on Urban Affairs 2000 (2000) 230-235
[Is Segregation Bad for Your Health? The Case of Low Birth Weight]
David M. Cutler: Ingrid Gould Ellen has written a fascinating paper on the impact of segregation on birth outcomes. Ellen shows clearly and convincingly that blacks living in more segregated cities have worse birth outcomes than do blacks living in less segregated cities. This is not true for whites. She also shows that the causality runs from segregation to outcomes and not the reverse.
The methodology that Ellen uses is similar to the one that Edward Glaeser and I employed in our study of the impact of segregation on economic outcomes for blacks. 1 Thus it is no surprise that I find it persuasive. Ellen's paper explores a new dimension of outcomes, however, and makes substantial headway. I take her finding to be true and robust: segregation is harmful to the health of black babies.
The central question raised by these results is why this is the case. The question is posed in the paper and speculated about, but it is never really resolved. Ellen shows that segregation matters because it is correlated with behavioral decisions that women make--both socioeconomic behaviors (education, marital status) and health behaviors (smoking, drinking). This finding echoes that of Ellen Meara, who has shown in a recent paper that less-educated women smoke more than more-educated women, and this explains a significant share of why low-birth-weight infants are more common among less educated women. 2
But why does segregation matter for these behaviors? The paper is unable to answer this question with the data available. I'd like to raise a few theories and discuss how they might be tested.
Income theory. The first theory is that segregation affects health because higher-income women can afford better health behaviors. Less-educated women may not be able to buy necessary food, medical care, and other services, [End Page 230] and this may adversely affect fetal development. This theory would not explain why segregation leads to increased smoking and drinking, but it could explain the effect of segregation through changes in educational attainment.
One way to test this theory is to examine the consumption patterns of women by income or education, matched with data on where they live. This has not been done, but other evidence suggests that this theory is not the complete explanation. National data show that the incidence of low birth weight among most Hispanic and Asian women is very similar to that of non-Hispanic whites, despite much lower income among these groups. In fact, the incidence is slightly lower for Mexican and Chinese mothers. 3 Thus it is clear that income by itself is not the explanatory factor.
Peer effects. A second theory is that segregation matters for birth outcomes because in more-segregated areas some women have worse behaviors and this is imitated by other women in a contagious way. Such peer effects may be bad (smoking, drinking) or good (vitamin intake, prenatal care).
I do not have a strong sense about whether this view is correct. Behaviors certainly do differ by racial group (although not always negatively--black women smoke less than white women). But whether this results from peer effects or from other factors that similarly influence women from different racial groups is unknown. I consider this a major research question.
Time allocation. The third theory is that segregation matters for health because women in more segregated areas have to spend more time on basic aspects of living and have less time for appropriate medical or nonmedical inputs to health. Depictions of life for poor mothers frequently stress the great complications of seemingly simple tasks. Taking the children to day care may involve a half hour or hour each way. Commuting to work or shopping, which are often not near the home, may involve similar amounts of time. With such a large time allocation devoted to these activities, it may be difficult for women to find time for doctors' visits or to ensure adequate rest and...