Using large samples of birth records, we reach four key conclusions:
1. School entry policies have large effects on schooling at motherhood: one-fourth of young Texas mothers born after the school entry date have a year less education than they otherwise would, had they been born before the entry date. For California, our estimate is one-seventh.
2. Education does not significantly impact fertility: women born just before and after the school entry date are equally likely to become mothers and give birth at similar ages.
3. Education improves mating market outcomes: women born just after the entry date have younger and less educated mates than women born just before.
4. Education does not significantly impact observable inputs to infant health and has generally small, but possibly heterogeneous, effects on infant health: women born just before and after the entry date have similar prenatal behaviors, as proxied by rates of smoking and prenatal care, and give birth to children of similar health, as proxied by birth weight, prematurity, and rate of infant mortality. There is some suggestive evidence of different effects of education on low birth weight by race and ethnicity.
Implementing our identification strategy requires information on date of birth, which is unavailable on most public-use files. We use an administrative data set on all births in California and Texas from 1989 to the present with information on mother’s date of birth and education, infant health, pregnancy behaviors (e.g., smoking and drinking), and paternal characteristics.2 These data allow us to focus contrasts narrowly around the school entry date, a challenge for earlier analyses in which either exact date of birth or large sample sizes were wanting (Angrist and Krueger 1991, Cascio and Lewis 2006).
A narrow focus on individuals born near the school entry date builds on the quarter of birth approach of Angrist and Krueger. First, it sidesteps the criticisms of Bound, Jaeger and Baker regarding seasonality of birth (assuming seasonal patterns are continuous at the school entry date). Second, it leads to a precise estimate of the relationship between within-year birth timing and educational attainment, circumventing statistical problems associated with weak instruments (Staiger and Stock 1997, Moreira 2003, Andrews, Moreira and Stock 2006).
The crucial assumption underlying this approach is that for dates near the school entry date, an individual’s date of birth is random. This assumption is plausible a priori, since parents are unlikely to strategically plan the exact date of birth of their child. Moreover, this assumption is testable—women born just before and after school entry dates should be similar in terms of predetermined, observable characteristics. We find that they are.