A note on "An Economic Approach to Abortion Demand." (response to Donna S. Rothstein, The American Economist, vol. 36, no. 1, p. 53)

Sun, Wei

This note re-estimates Ms. Donna S. Rothstein's abortion demand model using a continuous abortion price variable instead of dummy variable. The results suggest that the abortion price and Medicaid funding have insignificant effects on demand for abortion, which are different from Rothstein's. The policy implication of the results are also discussed.

Rothstein analyzed the socioeconomic factors affecting the demand for abortion. A cross-section of the 50 states and Washington D.C. for the year 1985 was examined. The dependent variable was the percentage of pregnancies of women aged 15-44 that are terminated by legal abortions (ARATE). Rothstein found that the average cost of abortions (PRICE) and unemployment rate (UNEMP) had significant negative effects on the demand for abortion. The disposable personal per capita income (DPINC), the availability of Medicaid funding for abortion (MEDST), the percentage of unmarried women age 15 and older (SINGL), the states that are located in the far west (WESTV), and the divorce rate (DRATE) had significant positive effects on the demand for abortion. The last variable in the model was the percentage of women aged 15 and over with 12 years of education or more (HIGHS), which had a insignificant effect on abortion demand. The specific results of her model were the following: (t-statistics in the parentheses):

ARATE = -11.654 - 0.041PRICE (0.993) (1.714) + 0.002DPINC + 3.837MEDST (4.027) (2.722) + 0.852SINGL + 6.140WESTV (5.833) (3.285) - 0.993UNEMP + 7.429DRATE (3.312) (2.200) -0.165HIGHS. (1.517)

[R.sup.2] = 87.4%.

Rothstein did not use a continuous price variable. Instead, she defined a dummy variable that equals one if the price is greater than $213 and zero if otherwise. (The national average price of abortion in 1985 was $213). The reason for the transformation from continuous price data to a dummy variable is unclear, and the unusual measure of price data makes the interpretation of the coefficient very difficult. For example, reducing the price in a state from $500 to $220 would have no impact in her model, but reducing the price from $214 to $212 would increase the demand for abortion.

The average price for each state in 1985 is available from the Alan Guttmacher Institute. The reestimation of the model with the continuous price variable (CPRICE) finds that most of the results are similar to Rothstein's. That is, the disposable per capita income, percentage of single women, and divorce rate are significantly positively related to abortion demand, while the unemployment rate has significant negative effect on abortion demand. But the effect of average price on abortion is much smaller and statistically insignificant. This result does not coincide with the conclusion made by Rothstein that the average price of abortion has a significantly negative effect on the demand for abortion. Also the coefficient for Medicaid funding (MEDST) is insignificant at the five percent level. These differences result solely from different measurement of price data. The results of the re-estimated model are as follows (t statistics in the parentheses):

ARATE = - 51.560 - 0.0004CPRICE (-4.541) (-0.027) + 0.0015DPINC + 1.447SINGL (3.138) (9 439) + 2.545MEDST + 6.722WESTV (1.674) (3.242) - 0.785UNEMPO + 9.371DRATE (-2.360) (2.642) - 0.057HIGHS. (-0.539)

[R.sup.2] = 988.6%

To rule out the possibility that multicollinearity between price and Medicaid funding could explain this result, an F-test or the joint significance of CPRICE and MEDST was constructed. The null hypothesis that CPRICE and MEDST jointly have no effect on ARATE is tested against the alternative hypothesis that at least one of the two coefficients is non-zero. The [F.sub.2, 42] equals 1.43, which is less than the five percent critical value of 3.23. We cannot reject the null hypothesis that price and Medicaid funding have no effect on abortion demand.

The conclusion that price of abortion has no effect upon the demand for abortion is not entirely surprising. First, the cost of an abortion is not high when compared with the expenditures of childbirth and child rearing. Second, abortion has essentially no substitutes. If a woman is pregnant and does not want to bring the pregnancy to term, she has no choice but to obtain an abortion.

The effect of price and Medicaid funding on abortion demand is obviously an important issue for both sides in the Medicaid funding debate. Much of the debate seems to assume that expanding Medicaid funding will cause much more women to obtain abortions. However, the results of this re-estimated model suggest that the price of abortion and the Medicaid funding for abortion have little or no impact on the number of abortions. As a result, Clinton's plan to expand Federal financing of abortions for poor women would not cause the number of abortions to rise. On the other hand, banning the use of federal and state funds for abortions would also not bring about a dramatic decrease in the number of abortions.

References:

Abortion Services in the United States, Each State & Metropolitan Area, 1984-1985.

S. Henshaw and J. Van Vort. New York: The Alan Guttmacher Institute, 1988.

Rothstein, Donna S. "An Economic Approach To Abortion Demand." The American Economist. Spring 1992. vol. 36. no. 1. 53 64.

Wei Sun, Department of Economics, University of Notre Dame, Notre Dame, IN 46556, 219/273-1870.