Factors Associated with the Timing of Abortions

Extensive literatures have investigated factors related to the abortion rate and the relationship between health and the economy, yet little is known about how economic fluctuations are associated with the abortion rate. There are a number of channels through which the business cycle could influence the number of abortions. Economic downturns may create doubts for some women regarding their financial ability to raise a child and thus make it more likely that they choose to have an abortion. Conversely, a stronger economy may lead to riskier behaviors that then lead to a higher rate of unplanned pregnancies and resulting abortions. Economic conditions may also impact the ability of women to afford an abortion.

This paper investigates the relationship between abortion rates and economic fluctuations using a sample of state-year observations during the period 1991 through 2012. The outcome of interest is the abortion rate and is based on annual reports by the U.S. Centers for Disease Control. In addition to the overall rate, rates by age and race/ethnicity are also employed as dependent variables. The explanatory variable of interest is the measure of state economic activity. The primary variable used is the state unemployment rate. To investigate potential heterogeneities in the relationship by sub-population, unemployment rates by age and race/ethnicity are employed in regressions where the corresponding abortion rate is used. Secondary regressions are estimated in which per-capita income is used as the indicator of state economic activity. While income is susceptible to autocorrelation, it can provide another perspective on the economic environment. We follow previous studies and include additional control variables including regulations on Medicaid funding of abortions and demographic characteristics.

The regression analysis employs state and year fixed effects to control for time-invariant state characteristics and year-specific effects common to all states. To account for heterogeneity in state size, the observations are weighted by population. Errors are clustered by state to account for potential within-state correlation.

Our preliminary findings indicate a significant, procyclical relationship between the abortion rate and the state unemployment rate. A one percentage point increase in the unemployment rate is associated with a roughly two percent decrease in the abortion rate. Based on the state average number of abortions, this one percentage point increase is associated with roughly 300 additional abortions.

Our findings potentially shed light onto the factors associated with abortions. While the decisions made by women are obviously very personal and idiosyncratic, our estimates suggest that economic factors may play a role, perhaps indirectly, in the decision. However, our analysis is subject to several limitations. The abortion data we employ is not available for all states and reports the number of abortions by state of occurrence rather than state of residence. Also, the interpretation of the results may be subject to the ecological fallacy given we attempt to infer about individual behavior based on group-level data.