How to Actually Reduce the Abortion Rate
Most "pro-life" policies are anything but. Pro-life and pro-choice people can come together on contraception, sex-ed, and racial justice.
Six months ago, I wrote about a curious result: in the United States, the abortion rate falls faster under Democratic presidents than under Republican ones. This is a little bit surprising and a little bit chastening, but at least for the audience I was hoping to reach I think it should be a source of relief! If you, like so many Christians, find yourself aligned Democratic policies but are uncomfortable with "the abortion stuff", I sincerely hope this lets you vote with a clear conscience and a joyful heart.
This time I would like to take that one step further, with much more thorough research to back me up. Pro-life advocacy, at least in the US, tends to be of the "culture war" sort, where the only way forward is to "defeat" the pro-choice side by legal warfare, shouted protests outside clinics, and shaming women who seek abortions. Too often a movement hoping to "shine Christ's light to the world" becomes cruel and self-defeating.
There are, of course, other ways to reduce the abortion rate. We could support the spread of contraception or comprehensive sex ed, or we could fight systemic racism or misguided global gag rules. In many of these ventures, we would find that the goals of the pro-choice and pro-life movements align, reducing the number of abortions while increasing the autonomy and welfare of the women in question. This would be a genuinely good thing -- if everybody is on the same side, less time and money needs to be spent "defeating" one's opponents, leaving more to make the world better for all.
And yet so often I see well-meaning pro-life friends dismiss this approach as "second best" or "giving in to the world", as if the measure of Christian success is how many debates we "won" and how many "opponents" we angered. My goal in this post is to point out that, according to a wide array of research, the opposite is true: culture-war approaches have fairly small effects on abortion rates compared to contraception/sex ed approaches. Fighting systemic racism is hard to quantify, but there is strong evidence that doing so would meaningfully reduce the abortion rate.
For the sake of readability, I will summarize the general evidence I've seen for various approaches in the literature on top, and then down below I will give a partial bibliography for some papers these estimates come from.
As a point of reference for understanding how big the effects are: in 2017, the US recorded about 13.5 abortions per thousand women ages 15-44 (hereafter "per 1kw"). In a typical year under a Republican president, the US abortion rate falls by about .26 per 1kw, and in a typical year under a Democratic president, the US abortion rate falls by about .64 per 1kw.
(Put another way: if these trends were to continue we'd need about 21 years under a Democratic president or about 52 years under a Republican president for the abortion rate to reach zero.)
Culture-War Approaches:
Currently, the most popular kind of pro-life legislation is to implement restrictions on legislation at the state level. Some of these have the goal of closing abortion clinics (e.g. requiring admitting privileges at local hospitals), while others try to change women's choices directly, either by requiring minors to get parental permission for abortions or by requiring doctors to straight-up lie about invented "risks" of the procedure. Studies typically find that these have an effect of between 0 and -2 abortions per 1kw. This is almost certainly an overestimate, since this kind of study looks at abortion rates within a state, and women can travel between states. The only study I've seen that tries to quantify the overestimate found that the true effect was about a third of the measured effect, so we can reasonably assume a typical state-level law has an effect between 0 and -1 at most.
The ultimate goal, one assumes, is to ban abortion entirely. The effects of this on the abortion rate are fairly unclear (for the obvious reason that measuring the number of abortions in countries where abortion is illegal is very difficult), but there isn't strong evidence that this would reduce the abortion rate by any meaningful amount. Today abortion rates in countries that have banned abortion seem to be about the same (actually a little higher) than countries where it is legal, and past evidence from Romania indicates that an abortion ban only affected the true abortion rate for a couple of years. In a world where abortion pills are (or soon will be) easy to buy on the internet, it is difficult to imagine this will have a significant effect. (This last consideration should also reduce our estimate of the effectiveness of state-level laws.)
The weirdest part of culture-war pro-life activism is the Mexico City policy, which (typically invoked under Republican presidents) revokes international family planning funding from organizations that mention abortion as an option. Studies repeatedly show that the policy significantly increases abortion rates in every affected country, along with increasing avoidable deaths from things like HIV. In theory, this should be a policy that everybody in the world opposes, because it doesn't seem to have any upsides.
Contraception and Sex Ed
By far the policy with the biggest known effect on abortion rates is improving access to contraception. The Affordable Care Act did exactly this, with a variety of results -- the contraception mandate decreased abortion rates by around 1 per 1kw overall (or by 2 per 1kw for women ages 20-29!), while the requirement that kids can stay on their parents' insurance until age 26 led to a whopping 2.58 - 4.05 per 1kw decrease in affected women! Other studies find similar effect sizes for other contraception-related interventions, some as large as 7 per 1kw. In any case, contraception use seems to explain the majority of between-population abortion rate differences, so this is a really big deal. Something like "Medicare for All" could have unbelievably large effects, but I'm not sure the pro-life movement is ready for this take yet.
The evidence on sexual education is a little bit more uncertain, but I think still fairly strong. There are, in general, two methods of sex ed in the US: "abstinence-only" sex ed focuses on teaching kids not to have sex as the only way to avoid pregnancy/STDs, as well as explaining failure modes in certain contraceptives. "Comprehensive" sexual education, on the other hand, includes education on how to use contraception (and in many cases more information on what sex is, consent, etc.) The general pro-life consensus seems to be that "abstinence-only" education is better, because comprehensive sex ed might lead kids to have more sex and that would be bad.
The research is pretty clear on a couple of points: comprehensive sex ed does not lead to significantly more sex than abstinence-only sex ed, but it does lead to significantly fewer unwanted teen pregnancies (like 40% fewer -- it's a huge effect), mostly because people use protection. It's not clear how well this scales up to the level of government policy (e.g. if I have a really well designed sex ed program, the teachers may just ignore it and say what they feel like), and so the estimates of effects on abortion rates range from 0 to 4 per 1kw. This creates room for pro-life activism not just at the state or federal level for comprehensive sexual education, but also at the local level to make sure teenagers are really getting the education that the law might say they are on paper.
Racism
This one is harder to quantify, but it's important: in 2008, non-Hispanic white women had about 12 abortions per 1kw, while non-Hispanic black women had about 40. Pro-life groups tend to be aware of this fact, citing it as a sign that the abortion industry is racist, usually with some references to Margaret Sanger (whose views on race are genuinely horrifying, but not super relevant to today's racial disparities in abortion rates.)
If we want to understand this disparity, instead of just using it as a cudgel, we should look carefully at what leads women to seek abortions and how these circumstances differ by race. People who have done this report a variety of factors, including discrimination and unequal access to the medical system (in particular, less access to contraception, less education on which forms of contraception are effective, less of access to family-planning / ob-gyn services, and mistreatment by those services when they are accessible) and strong neighborhood effects on "risky" sexual behavior (there's a strong literature linking these to generational poverty, mass incarceration, and even direct links back to slavery. This is the sort of thing where racism is affecting every single part of the system, leading to disparities building on disparities.).
I don't think anyone knows "the way to solve racism", but the evidence is clear that a significant chunk of the abortion rate is driven by systemic factors, and a movement that pretends to be pro-life without addressing the intersecting effects of racism and poverty will continue to be deeply ineffective.
Other Contributors
There are of course other contributors to the abortion rate and possible interventions. I currently think that "convincing people to become pro-life" has essentially no effect on abortion rates, but I'm not certain enough of this to write a detailed position. Similarly, I would expect anti-poverty interventions to have large effects, but the literature is complicated and I'm not sure I'm comfortable making a quantitative claim yet.
In summary, contraception and sex-ed are not just liberal compromise positions. They are genuinely pro-life, probably more so than the majority of current pro-life initiatives. The pro-life movement should unreservedly back them and find common ground with the pro-choice movement for the betterment of all.
Appendix: here is a list of papers with 1-2 sentence descriptions I made for myself while I was exploring the topic. It is incomplete (missing papers I was unimpressed by and also missing papers I forgot to write down.) I have not edited it, so be warned. Any errors are my fault. I might add to this over time because I think it's helpful as my views develop to have a list to consult.
Econometric note: the vast majority of these use either OLS or DID, and so constitute weak evidence at best. I tried to form my views based on multiple papers measuring different things, which I think makes the evidence a bit stronger. There is rampant misunderstanding of p-values throughout this part of the literature, so be warned. Given that an experiment would be unethical and I can't think of a reasonable instrument, this may be the best kind of evidence we can expect.
Contraception
"Planning parenthood: the Affordable Care Act young adult provision and pathways to fertility" (Abramowitz 2017) uses a DID-style analysis to estimate the effect of Obamacare's "children can stay on their parents' insurance until age 26" mandate on abortion rates. (The theory of change being that the mothers now have access both to more contraception and better services if they were to have the baby). Huge effect size -- between -2.58 and -4.05 per 1kw on affected women.
"Contraception Use, Abortions, and Births: The Effect of Insurance Mandates" (Mulligan 2015) examines the effect of the Affordable Care Act's insurance mandates on contraception use and abortion rate, comparing states who previously had contraception mandates to states that did not. Potentially an interaction issue with states that also enforced anti-abortion laws around the same time? Finds an overall affect of around -1 per 1kw, with much larger effects (~-2 per 1kw) on women age 20-29.
"Preventing Unintended Pregnancies by Providing No-Cost Contraception" (Peipert, Madden, Allsworth, and Secura 2014) studies the effect of providing free long-acting reversible contraception (alongside other contraception methods) on abortion rates, comparing an experiment group in St Louis to control groups in nearby areas. (Part of their goal: contraceptions with low failure rates compared to "the pill" and condoms. They find a massive effect, around -7 per 1kw.
New has a critique of this study that hinged on a data-specific fact, I wasn't able to figure out who was right before I got bored.
"Long-Acting Reversible Contraception Free of Charge, Method Initiation, and Abortion Rates in Finland" (Gyllenberg et al. 2017) looks at a program to provide LARC free in Finland. Finland has a much lower abortion rate than the US (hovering around 1 per 1kw), but they find a significant decrease from 1.1 to .9 per 1kw. Can debate if the right way to interpret this in the US context would be -.2 per 1kw, or -16% (-2.16 per 1kw in US numbers), but the second seems more plausible.
"Advanced provision of emergency contraception does not reduce abortion rates" (Glasier et al 2006) looks at program in Lothian, Scotland providing free emergency contraception to women. Many women used the program, they do not find a significant effect. Table 4 shows a point estimate of around - 1.5, so this study doesn't contradict the others in this section -- it seems like part of the problem is just that this study is underpowered.
"Relationships Between Contraception and Abortion: A Review of the Evidence" (Marston and Cleland 2003) finds generally unclear relationships across countries' levels of contraception and abortion over time, particularly when fertility rates in general are rising quickly. Hard to say anything quantitative.
"Declines in Unintended Pregnancy in the United States, 2008–2011" (Finer and Zolna 2016)
"Abortion Rate and Contraceptive Practices in Immigrant and Swedish Adolescents" (Helström, Zätterström, and Odlind 2006) finds that immigrant adolescent women have more abortions than native-born adolescent women in Sweden, and links this to higher rates of contraceptive use and counselling by Swedish women.
Effect of Legal vs. Illegal Abortion:
"Unintended pregnancy and abortion by income, region, and the legal status of abortion: estimates from a comprehensive model for 1990–2019" (Bearak et al 2020) finds that the abortion rates in countries where abortion is legal (about 40 per 1kw) are not significantly higher than those where abortion is illegal (about 40 per 1kw) or where abortion is permitted only to save the mother's life (about 36 per 1kw). If we exclude China and India from the "abortion legal" set, the abortion rate drops to 26 per 1kw. Abortion rate is falling fast in such countries, rising in other countries. Seems that socioeconomic factors are explaining much more of the abortion rate than a country's laws around abortion.
"Predicted changes in abortion access and incidence in a post-Roe world" (Myers, Jones, and Upadhyay 2019) looks at the effects of a Roe v Wade reversal, estimating a 12.8% decrease in the US abortion rate (about -1.72 per 1kw) This assumes no states that don't currently try to ban abortion will, that previous estimates from Texas are accurate, and also that women will not e.g. buy abortion pills online.
"Unsafe abortion: the preventable pandemic" (Grimes et al. 2006) explores a number of facts surrounding unsafe illegal abortions. For our purposes, figure 5 is interesting: when dictator Ceausescu bans contraception and abortion in 1966, birth rates essentially double, but they're close to where they started within five years. Seems like strong evidence that women have found access to illegal contraception or abortion -- the latter hypothesis is evidenced by the climbing death rate from illegal abortions. It's not clear at all that the law reduced the total number of abortions after the first handful of years, while it does appear to have killed a lot of women.
Effect of State-Level Laws:
“State abortion rates the impact of policies, providers, politics, demographics, and economic environment” (Blank, George, and London 1996) finds significant effect of laws banning Medicaid from paying for abortions (around 13% decrease in abortions performed within-state, or about -1.75 per 1kw in today's numbers). More importantly: find a much smaller effect on abortions performed on state residents (about 5% decrease, or about -.675 per 1kw in today's numbers). Means that any study (including most of the below) that doesn't keep track of women travelling to receive abortions is going to significantly overestimate the effect of a given policy.
"Assessing the Impact of Supply-Side Abortion Regulations in the United States" (Austin 2019) explores TRAP laws, conducting a literature review followed by a series of regressions. A variety of effects in the literature, including a .77% decrease in the abortion rate (about -.1 per 1kw today), a 13% decrease in the abortion rate (about -1.75 per 1kw today), a 5.5% decrease in the abortion ratio (not rate, this is abortions per 1000 births), and a very strong effect on late-term abortions. Austin's model itself finds estimates of a similar magnitude, but not statistically significant (-1.25 per 1kw for ASC, and +.57 per 1kw for admitting privileges laws)
“Analyzing the Effect of Anti-Abortion U.S. State Legislation in the Post-Casey Era” (New 2011) considers three types of state-level laws: bans on public funding (via medicaid), parental involvement laws (for minors seeking abortions), and "informed consent" laws (some of these require doctors to give information on e.g. adoption to women seeking abortions, others force doctors to lie to their patients about supposed "risks" of abortions.) Finds effect sizes of ~ -1.5, -0.5, and -1 per 1kw.
"A Critical Reexamination of the Effect of Antiabortion Legislation in the Post-Casey Era" (Medoff and Dennis 2014) points out an unbelievable number of coding errors on New's part, as well as some technical mistakes. They find similar (slightly smaller in magnitude) effect sizes, although the "informed consent" effect shrinks to -0.5 and is no longer statistically significant.
The biggest problem (which Medoff and Dennis point out, but I'm not convinced by their solution to) is that this is necessarily an overestimate: a correct approach would look at abortions by state of residence, but this sort of data is not available. Since 49% of abortions in Kansas, for example, are performed on out-of-state residents, and 20x as many Wyoming women seek abortions outside of Wyoming as within, this is a serious effect, and makes the effect of this sort of policy look bigger than it is. (At least, on number of abortions! Having to travel certainly has a big impact on the lives of the women seeking abortions, which is at least supposedly not the intended consequence.)
"Using natural experiments to analyze the impact of state legislation on the incidence of abortion" (New 2009) compares the abortion rates in state which have passed abortion restrictions to those whose restrictions have subsequently been overturned by a judge. Many of the 2011 paper's issues apply here too, but for what it's worth he finds an effect size of -1.24 for informed consent laws and -1.13 for parental involvement laws.
"The Determinants and Impact of State Abortion Restrictions" (Medoff 2002) runs a probit regression, and finds a positive (that is, increasing abortions) but statistically insignificant effect of state abortion restrictions.
"Changes in abortions and births and the Texas parental notification law" (Joyce, Kaestner, and Colman 2006) studies the effects of a parental notification law in Texas, and in particular its effect on teenage abortion rates. They find a decrease in abortions (but an increase in late abortions for teenagers who enter the second trimester after turning eighteen!). Their findings are consitent with the literature they cite, noting that "Results from previous studies have indicated that the abortion rate of minors decreases 13 to 42 percent after enforcement of a parental notification or consent law, if abortions are measured according to state of occurrence. Only a few studies have reported a decline associated with these laws when abortions are measured according to state of residence." Unclear how many of these minors turn to illegal abortion methods, since they do not find a corresponding increase in number of births.
"Using the Synthetic Control Method to Determine the Effect of Ultrasound Laws on State-Level Abortion Rates" (Gius 2019) uses a fancy econometric technique called "synthetic controls" to estimate the effect of state-level ultrasound laws. The SC method is consistent with no effect (he doesn't give a point estimate), while an older paper of Gius's using fixed effects found an effect of about 10.4% (about -1.4 per 1kw).
"How Far Is Too Far? New Evidence on Abortion Clinic Closures, Access, and Abortions" (Lindo, Myers, Schlosser, and Cunningham 2017) finds that distance to abortion clinics matters, in particular finding that a 25-mile increase in distance for those who currently live close leads to 10% fewer abortions (-1.35 per 1kw today).
Racism
"A Case Study of Race Differences Among Late Abortion Patients" (Lynxwiler and Wilson 1995) find that black women are more likely to seek late abortions than white women, and explores reasons why. Citing previous literature, they note that:
Black women obtain nearly twice as many abortions as white women
Black women are more likely to face unintended and accidental pregnancies than white women
All this despite the fact that black women are more pro-life than white women overall
Lynxwiler and Wilson find strong evidence that the gap in timing corresponds in part to different income levels among races, different levels of support for the woman's decision (women whose social structures do not support the decision to abort put it off, leading to more late-term abortions), and different levels of fear in telling others they've become pregnant.
"Disparities in Abortion Rates: A Public Health Approach" (Dehlendorf, Harris, and Weitz 2013) explores reasons for . They find:
Women of color and poor women are more likely to face unintendent pregnancies. These effects are independent.
Similar effect holds for initiating sex at a young age.
Similarly, poor women and women of color are less likely to use contraception, and more likely to use less effective contraception measures.
What causes this? They cite a variety of literature showing that high-risk sexual activity is highly correlated with the neighborhood in which a person lives, and that after controlling for neighborhood the link with race or socioeconomic status goes away. (There's a post-treatment issue here, so not entirely clear what to make of this.) Similarly, lack of insurance coverage causes less use of contraception, and is linked with SES and race independently, as well as discrimination within family-planning services and lack of access to real (i.e. non "abstinence-only" sex ed.)
"The Black Family and Mass Incarceration" (Western and Wildeman 2009) makes nuanced complex points -- black single parenthood, mass incarceration, the decline in inner city jobs, educational disparities, all closely related. Effects of issues dating back to slavery, but racism present every step of the way.
(I'm not going to list every paper on racism ever published here but the general consensus is that racism is a really really big deal and we ignore it at our peril.)
Sex Education
"Abstinence-Only-Until-Marriage: An Updated Review of U.S. Policies and Programs and Their Impact" (Santelli et al. 2017) reviews a variety of meta-analyses. Finds strong evidence across many gold-standard papers that abstinence-only education does not decrease sexual activity, while comprehensive sex ed programs do.
"An Economic Analysis of Abstinence-Only Sex Ed" (Collins 2011) explore the results of abstinence-only education in Alabama, finding increases in various STDs and pregnancy-related outcomes. Switching to an abstinence-only sex ed program increased abortion rates by between .7 and 1.2 per 1kw.
"Abstinence-only education and teen pregnancy rates: Why we need comprehensive sex education in the US" (Stanger-Hall and Hall 2011) finds little correlation between abstinence-only education and abortion rates across states. SUPER confounded by the link between abstinence-only education and conservative religious communities.
"The Impact of State Abortion Policies on the Abortion Rate" (Danel 2015) finds, with a more controlled regression, a significant impact of abstinence-only education on abortion rates, about .04 more abortions per pregnancy. (In our numbers this is about 4 per 1kw given today's pregnancy rates, but take this with a grain of salt because abstinence-only also significantly increases the number of pregnancies.) For what it's worth, Danel's estimates of other effects are similar to what we've seen (but insignificant b/c low power.)
"The Effects of State‐Mandated Abstinence‐Based Sex Education on Teen Health Outcomes" (Carr and Packham 2016) finds essentially no result of abstinence-only education on either pregnancy or abortion rates. They control for NARAL ratings (which could be affected post-treatment) and also include states that teach about contraception in their treatment group, which seems like it might shrink the results.
"Do State-Based Policies Have an Impact on Teen Birth Rates and Teen Abortion Rates in the United States?" (Chevrette and Abenhaim 2015) similarly finds no effect. Their regression model seemed kinda sketchy.
"Abstinence-Only and Comprehensive Sex Education and the Initiation of Sexual Activity and Teen Pregnancy" (Kohler, Manhart, and Lafferty 2008) runs an RCT comparing abstinence-only to a comprehensive sex-ed program, finding that the comprehensive program decreased teen pregnancy by about 60%. The abstinence-only program decreased it by about 30%, but not statistically significant. No evidence that the comprehensive teens had more sex than the abstinence-only teens.
The Mexico City Rule
"USA aid policy and induced abortion in sub-Saharan Africa: an analysis of the Mexico City Policy" (Brooks, Bendavid, and Miller 2019) finds that the Mexico City policy led to an increase of about .5 per 1kw in the abortion rate in affected countries, or about 40%.
"Contraceptive supply and fertility outcomes: evidence from Ghana" (Jones 2013) finds that the effect on abortion rates differs by location: urban women are not too affected, while rural abortion rates increase by about 2% (4% for the poorest women.)