Provider Availability, Race, and Abortion Demand.

• Author: Brown, Robert W.
• Position: Research information - Statistical Data Included

Robert W. Brown [*]

R. Todd Jewell [+]

Jeffrey J. Rous [++]

Variations in the availability of abortion providers may impact the demand for abortions since greater provider availability reduces the travel cost associated with obtaining an abortion. This paper applies a fertility-control model to estimate the responsiveness of abortion demand to travel-cost variations using individual data from all births and abortions of women over age 20 in the state of Texas for 1993. The probability that a pregnant woman chooses an abortion appears to be sensitive to availability-induced variations in the travel cost of abortion services. Controlling for the endogeneity of travel distance, the results suggest that pregnant women who reside in counties with longer travel distances to the nearest abortion provider have lower probabilities of aborting their pregnancies than women in counties closer to abortion providers. Simulations show that changes in travel distance will have relatively large impacts on overall abortion rates and, furthermore, that these effects vary across race. In addition, these simulations show substantial differences by race in the effects of changes in other explanatory variables.

1. Introduction

   In the 1973 case of Roe v. Wade, the U.S. Supreme Court ruled that states cannot prohibit a woman from having an abortion during the first trimester of pregnancy. However, states may regulate abortion services in the second trimester or prohibit abortions during the third trimester of pregnancy. The Court's decision does not mandate the provision of abortion services; therefore, it does allow for the availability of abortion providers to vary within a state. Since that landmark decision, which effectively broadened abortion availability in the United States, there has been a great deal of debate on the degree to which abortion demand is sensitive to the legal availability of abortion services.

   Variations in the availability of abortion providers impact the demand for abortions in several ways. For example, greater geographical availability of abortion providers reduces the travel cost associated with obtaining an abortion; if demand is sensitive to changes in the full cost of abortion services, then greater availability will increase the quantity of legal abortions demanded. In addition, the availability of abortion providers itself could signal social approval for the abortion decision, lowering the psychic cost of terminating an unwanted pregnancy. Finally, provider availability also creates a potential moral hazard problem in that knowing abortion services are available may produce an incentive for individuals to be less careful in using contraception to avoid unwanted pregnancies. In each of these cases, greater availability of abortion providers induces individuals to increase their consumption of abortion services.

   Empirical research has attempted to assess the direct impact of variations in the local availability of abortion services on abortion demand. Deyak and Smith (1976) measure the economic benefits of Roe v. Wade by estimating a travel-cost demand function for abortions by women of all ages in New York, a state with liberal abortion statutes prior to Roe v. Wade. The national liberalization of abortion providers following Roe v. Wade eliminated the need for out-of-state women to obtain legal abortions in New York, reducing the necessary travel cost of having an abortion. Using their travel-cost demand estimates, the authors predict the gain in consumer surplus resulting from greater availability of providers in other states. Over a 10-year period prior to Roe v. Wade, the forgone benefits of liberalized abortion statutes are estimated to be between $12 million and $30 million.

   Medoff (1988) estimates the demand for abortions by women of childbearing age in the United States as a function of the dollar cost of abortion services and other socioeconomic factors impacting demand. Demand is shown to be responsive to changes in the dollar cost of provider services; however, these data do not allow for testing the effect of variations in provider availability on abortion demand. Joyce and Grossman (1990) use micro-level data from women in New York City to investigate the demand for prenatal care. The authors estimate an abortion equation to control for self-selection of birth outcome and find that provider availability significantly increases the probability that a pregnancy ends in abortion for their sample of white women. However, they also find that availability has a significantly negative effect on the likelihood of an abortion for blacks and an insignificantly negative effect for Hispanics. [1]

   Gohmann and Ohsfeldt (1993) estimate the demand for abortions using pooled data on U.S. states and include a state-level variable to index provider availability, which proves to be a statistically insignificant determinant of both abortion and pregnancy rates. Meier et al. (1996) examine 23 state-level restrictions and find no evidence of any impact on state-level abortion rates. As the authors point out, however, the variation in the number of providers is a potentially important determinant of abortion rates. Garbacz (1990) also uses state-level data and finds that the number of abortion providers in rural areas has a significantly positive effect on the abortion rate. Brown and Jewell (1996) use county-level data to show that local abortion availability can affect both abortion and pregnancy rates. The authors find that counties with abortion providers have higher abortion and pregnancy rates. Powell-Griner and Trent (1987) find that women residing in metropolitan areas tend to have more abortions; the au thors attribute this finding to the greater availability of abortion services in metropolitan areas.

   Some studies have concentrated on the abortion decision of younger women. Results from both Lundberg and Plotnick (1995) and Haas-Wilson (1996) suggest that teens who live in states with greater abortion availability are more likely to have abortions. Ellertson (1997) finds that teens residing in states with parental consent laws are more likely to travel out of state to receive abortions. Currie, Nixon, and Cole (1996) use data from the National Longitudinal Study of Youth and find that county-level abortion access has an insignificant effect on the outcome of a pregnancy. In a related study, Kane and Staiger (1996) use county-level data from U.S. states to show that local abortion availability affects the teen birthrate; however, the authors do not estimate the relationship between local abortion availability and the teen abortion rate.

   None of these studies controls for the possible endogeneity of provider location. Specifically, abortion provider location and, therefore, measures of access such as travel distance to a provider or provider density in a geographical area may be determined by many of the same unobserved factors that also contribute to a woman's abortion decision. For example, a higher level of "right to life" group activity may not only have a negative effect on the demand for abortion services, but it may also have a positive effect on travel distance if their activity is successful in closing down clinics. Alternately, psychic cost may have a negative effect on abortion demand and be negatively correlated with travel distance if women are more likely to be seen at a local provider by friends, relatives, and so on. Failing to control for such correlation could seriously bias the estimation results. Because the factors causing the correlation are unobserved, it is impossible to predict the direction of the bias.

   This paper applies a model of fertility control to directly estimate the responsiveness of abortion demand to variations in the travel-cost component of the full cost of abortion services. We utilize a two-stage estimation method to control for the potential endogeneity of abortion availability. Data on the availability of abortion providers in the 254 counties in Texas, 235 of which were without an abortion provider as of 1992, provide a unique opportunity to approximate the travel cost of abortion services for Texas residents. [2] This information is supplemented with individual data on pregnancy outcomes for 1993. Unlike most previous literature, these Texas data allow for comparing the fertility-control decisions among large samples of black, Hispanic, and white women. The estimates suggest that a woman's decision to abort a pregnancy is sensitive to availability-induced variations in travel cost. We find that women residing in Texas counties with longer travel distances to the nearest abortion provider h ave lower probabilities of aborting a pregnancy; in addition, the magnitude of the travel-cost effect differs substantially across race. [3]

2. Methodology and Data

   Michael (1973) models fertility control as a household decision made on the basis of the cost and benefit associated with having an additional child over time. A household's net benefit from an additional child reflects the effective excess demand for children at given prices, wealth constraints, time constraints, preferences, and household production of substitutes and complements for children. If the net benefit of an additional child is negative, then it is predicted that a household will practice fertility control in order to reduce the probability of birth. In this context, abortion is considered as a means of fertility control. Accordingly, the abortion decision is a function of household income, preferences, the opportunity cost of an additional child, and the full cost of abortion services, including the dollar cost plus the travel cost and any psychic cost. Within this framework, we focus on the impact that variations in the travel cost of abortion services have on the individual demand for abortions in Texas. Expanded availability of abortion services lowers the travel-cost component of...