Determinants of teenage birth rates as an unpooled sample: age matters for socioeconomic predictors

American Journal of Economics and Sociology, The, Jan, 1999 by Annette Tomal

I

Introduction

United States teenage pregnancy and birth rates, although declining, are still among the highest in the industrialized world (Spitz, 1996). Teen mothers gave birth to 13 percent of the babies born in 1995 and account for about one-third of unwed mothers (Conrad, 1996). Most econometric analyses of teenage birth and abortion rates, however, view teenagers as a pooled sample (Haas-Wilson, 1996; Meier, et al., 1996; King, et al., 1992), even though characteristics or determinants of teen pregnancy and birth may differ significantly for older versus younger teens.

Teens under fifteen years old, for example, accounted for only 3 percent of all teen pregnancies and abortions in 1990 (Spitz, 1996). Also, eighteen- to nineteen-year-old pregnant teens face substantially different opportunity costs than younger teens; they are approaching or past high-school graduation and are also at an age where motherhood may be more socially acceptable, particularly if combined with marriage.

Much of the research in the area of teen pregnancy and birth has focused on the long-term costs to teen mothers and their babies, such as chronic unemployment, low-paying jobs, and lack of a high-school diploma. Researchers now recognize that these costs may be incurred by many of these teen mothers even if they had never become pregnant, primarily because many teen mothers come from socioeconomically disadvantaged backgrounds (Hoffman, 1993).

Applying rational choice theory to the debate of the cause-effect relationship between poverty and teen pregnancy and birth, Conrad (1996) theorized that "many unmarried teens are having babies as a rational response to prevailing economic conditions - specifically the job market they face." A study analyzing teen pregnancy in England and Wales and the effects of social policy from 1951 to 1992 concluded that education and job opportunities hold more promise for curtailing the birth rate than removal of welfare and housing benefits (Selman, 1994). Leibowitz and colleagues (1986) hypothesized that teens have a high time preference for the present and are less likely to carry a pregnancy to term if they are in high school and earning good grades; Ribar (1994), however, asserts that teen childbearing is an endogenous determinant of high school completion.

Other researchers dispute that teenagers make a rational choice about pregnancy and birth. Hymowitz (1994) believes that intact families are an important deterrent to teen pregnancy, and Furstenberg (1990) believes that a live-in father is of particular importance. Other critics cite poor sex education as a primary cause for teen pregnancy (Calderone, 1988).

Many analyses use state-level data (Medoff, 1988; Rothstein, 1992) and can thus measure the effects of differing abortion parental involvement laws, government abortion funding, and state welfare and family planning policies. However, for economic and demographic variables such as income level, unemployment rate, minority populations, and population density, the use of state-level data does not recognize substantial differences that may exist in smaller regional areas that could bias the coefficients and significance levels of these state-level determinants.

II

Model and Estimation

This study econometrically analyzes teenage birth rates as both a pooled sample and as an unpooled sample by separating teenagers into two age groups - under eighteen and eighteen to nineteen years old - to determine not only the importance of various predictors on teen birth rates but also to determine whether the effects are statistically significantly different for the two age groups. The independent variables attempt to model both actual and perceived economic environments and opportunities faced by teenagers. The model uses county-level data from one state, which may limit generalizability to some extent. However, the use of county-level data focuses primarily on whether birth rate determinants for the two groups of teens are statistically significantly different, while also attempting to control for the effects of state-level abortion funding policies, state welfare and family planning policies, and abortion parental involvement laws. Border crossing may, of course, affect some of the counties' abortion rates, and therefore also birth rates.

A regression equation is estimated for (1) all teens, (2) teens under eighteen, and (3) teens eighteen to nineteen years old:

Birth Rate = f(population density; education level; per capita income; extent of poverty; unemployment rate; racial composition; family stability)

The birth rate is defined as the number of births per estimated thousand females in the age group being analyzed. The regression equation is estimated for (1) all counties and (2) all counties with greater than 80 percent white population.

POPDENS (population per square mile) measures the extent of urbanization, which may reflect a greater variety of economic opportunities and also greater availability of abortion services. In fact, living in an urban area does contribute to a greater probability of abortion (Currie, 1996; Liu, 1995). %WHITE (percent of county's residents who are white) addresses the empirically higher probability of nonwhite teens' births (Ventura, 1994; Spitz, 1996).