Birth weight and infant health for multiple births

I provide plausibly causal estimates of the effect of birth weight using data from the universe of twins, triplets, and quadruplets born between 1995 and 2000 in the United States. Infants from higher-order multiple births have lower birth weights, but experience smaller reductions in infant health when they are low birth weight. OLS estimates using a rich set of controls show that this result holds when comparing singletons to multiples and when analyzing historical data from before the widespread use of in vitro fertilization. These results suggest that the paradox of low birth weight remains even after controlling for unobserved maternal heterogeneity.