In the continuation of this research project, we propose to: 1. Estimate the effect of teenage childbearing in the U.S. on socioeconomic and biological outcomes of the children, including the prevalence of low birth weight, pre-term delivery, infant mortality, health and behavioral problems, slow growth and development, and poor access to health care, using statistically robust, non-parametric bounds developed in the initial funding period of this grant. 2. Replicate and validate, using alternative data sources, work done in the previous funding period, examining the effect of teenage childbearing on maternal outcomes, including post-high school educational attainment, labor force participation rates and wages achieved, the propensity to marry and remain married, the propensity to receive welfare, and the propensity to live in poverty. we will further expand this research by considering maternal outcomes from the child's perspective, i.e., at different ages of the child's life, to determine the potential impact of early childbearing on the development of the offspring of teen mothers. 3. Further develop and test the validity of the evaluation design developed in the previous funding period of using the occurrence of random (chromosomally aberrant) miscarriages among teen women to form a natural experiment with which to identify the causal effect of the failure of teen mothers to delay their childbearing. In this analysis, we will explore strategies for dealing with underreporting of fertility events such as pregnancies, abortions, and miscarriages in establishing our bounds by incorporating national health statistics on these events into our estimation. 4. Test economic and sociological theories, as well as statistical methods, of the decisions of some women to have children as teenagers by validating whether the estimators associated with these models yield estimates of teenage childbearing that lie within our robust bounds. For example, we will test parametric specifications to deal with the endogeneity of teenage childbearing that is frequently used in the literature, e.g., family (sibling) fixed effects and linear regression controls for background effects, against our robust bounds. We will also explore ways to tighten our proposed non-parametric bounds by using auxiliary information from medical research and by using behavioral assumptions, such as comparative advantage and common factor structures, in an empirical specification less highly parameterized than in the current literature.
