Do pro-life laws work?

How many abortions occur when abortion is illegal?

Does welfare reduce abortion?

Does sex education reduce abortion?

Does contraception reduce abortion?

Democrat Presidents reduce abortion rates more than Republican Presidents

Do pro-life laws work?

There is a common claim that banning abortion is pointless because it doesn’t work. Women will get abortions anyway; indeed, they will get unsafe abortions and die as a result. I deal with the latter claim under ‘Does legalising abortion prevent women dying from backstreet abortions?’

It is worth noting that this same argument was used in support of the slave trade. It was argued, for example, that keeping the slave trade legal would allow it to be better regulated, and hence allowing better conditions for the slaves on ships, for example. Abolitionists were often blamed for uprisings in which many people were killed, because they had allegedly stoked tensions by giving slaves the hope of freedom. All of us find the pro-slavery arguments unconvincing for two reasons: a) even if they were empirically correct, the dignity of human beings kidnapped and trafficked into slavery deserved to be recognised in law; and b) it is profoundly implausible that the empirical claims are correct.

The same is true in this case. Take the empirical claims, for example. In fact, there is no serious controversy about whether giving legal protection to unborn children reduces abortion rates. Although some otherwise credible people sometimes repeat this claim, there is in fact no evidence for it at all. By contrast, there is overwhelming evidence that pro-life laws do work, even though they don’t entirely eliminate abortion (just as no laws entirely eliminate the crime).

It would be somewhat tedious to list all the evidence that pro-life laws reduce abortion rates here, though I am working on a paper to that effect. For those who are sceptical, Secular Pro Life have put together long lists of studies supporting this conclusion here and here. Philip Levine’s Princeton University Press book Sex and Consequences is also highly recommended.

But for perhaps the most direct and clear evidence, consider the Turnaway study, where women seek abortions, are denied because they are over the gestation limit, and go on to either have an abortion elsewhere (illegally or out of state), miscarry, or have the baby. This study provided incontrovertible evidence that pro-life laws work: of women turned away from the clinic, over 2/3 went on to have the baby, with most raising the child themselves and a small minority placing the child for adoption. Moreover, a variety of such ‘turnaway’ studies have been done historically and recently. The lead author of the study – a passionate pro-choice advocate – wrote an article summarising some of the evidence that pro-life laws work and pleading with fellow pro-choicers not to keep making this claim.

The only evidence usually cited for the claim that pro-life laws don’t work is a couple of studies in The Lancet which show that countries with pro-life laws have similar abortion rates to those with pro-abortion laws. Some of the problems with drawing such conclusions from these studies are, to be candid, obvious. Others are less so. In short, the problems are:

1. The methods used to estimate abortion incidence in pro-life countries unreliable, for obvious reasons. The standard method used for estimation, the Abortion Incidence Complication Method, is critically analysed under ‘How many abortions occur when abortion is illegal?’
2. Secondly, the studies do not control for any possible confounders, and hence insofar as they are interpreted to claim that there is no causal relationship, they commit the correlation-causation fallacy. There are clearly many other differences between pro-life and pro-choice countries which could explain higher abortion rates in the former – for example, much higher rates of poverty and much lower access to reliable contraception. When like countries are compared with like (as with the UK and Ireland/Northern Ireland), it is clear that pro-life countries have far lower abortion rates. This is a very obvious and fatal problem with the conclusion, which would probably be noticed by an astute child.
3. Finally, the studies themselves point to a more salient statistic: the proportion of pregnancies ending in abortion was much higher on continents with primarily pro-choice laws. For example, the most recent study explicitly notes: “In the group of countries where abortion is prohibited altogether by law or allowed only to save a woman’s life, 48% of unintended pregnancies ended in abortion… this proportion was substantially higher, at 69%, in countries where abortion is allowed on request”. The same trend is true when all pregnancies are considered, not only unintended pregnancies.

Hence there is no remotely persuasive evidence that pro-life laws don’t work, and an abundance of evidence that they do. There is no reasonable controversy on this question.

How many abortions occur when abortion is illegal?

Obviously, no one knows for sure. Abortions are generally massively underreported when women are asked directly, even when entirely legal and more so when abortion is illegal. But this is an important question for which we should understand the evidence commonly cited, since claims about illegal abortions are often used to argue for abortion legalisation: for example, that abortions are very common anyway, or that abortions cause a huge economic burden on the healthcare system, or that many women die from illegal abortions (these arguments are dealt with separately).

There is one particularly common method used to estimate abortions where abortion is illegal: the Abortion Incidence Complication Method. This involves clinics and hospitals estimating how many women they see per month with post-abortion complications. Since many of these women are having miscarriages, an estimate is used of women obtaining medical care for miscarriage complications. This is based on (among other assumptions) the idea that women will only seek medical help for miscarriages after about 13 weeks of pregnancy, and that these are about 3.41% of live births. This figure is subtracted from the total number of estimated post-abortion complications to yield the complications due to illegal abortion specifically. Then, another estimate is given of the proportion of illegal abortions requiring medical treatment – if 20% of illegal abortions are thought to need medical care, then the estimated complications from induced abortion are multiplied by 5 to reach the total number of abortions. Most of these involve guesswork, as is often conceded in the studies themselves.

The logic is perfectly sound, but there are a number of critical flaws. These are comprehensively addressed in a forthcoming paper by a colleague (please ask for details), but here are a couple of the key issues:

1. The method relies on a very low estimate for miscarriages requiring medical attention – 3.41%. In fact, it is known that the number is significantly higher in countries such as the UK (up to 7.8%) and Ireland (7.8% from personal correspondence with the author, when late miscarriages are included). In many countries with pro-life laws, there are significantly more risk factors for miscarriage, such as ethnicity, malaria, malnutrition, being underweight, and HIV infection. A more plausible estimate of miscarriage prevalence alone radically reduces the estimate of abortion complications and abortions. For example, assuming 1 million live births a year, 100,000 ‘abortion’ complications presenting to hospital, 80% of miscarriage complications needing treatment actually getting treatment, and miscarriages needing treatment totalling 3.41% of live births, you get an estimate of 72,720 complications from induced abortion in hospital each year. Changing 3.41% to 7.8% yields an estimate of only 37,600 complications from induced abortion in hospital each year. Changing to 10%, given a likely higher rate of miscarriage in developing countries (10% being used for illustration), yields only 20,000. Hence the estimates depend critically on this figure, and 3.41% is an entirely implausible assumption.
2. This method therefore yields entirely implausible results. For example, applying this method to known data in the UK from 2002-2003 regarding miscarriage complications yields the result that 57% of women presenting to hospital in the UK with miscarriages are in fact presenting with complications of induced abortion – around 25,000 in fact. This is absurd in light of the fact that virtually none of these women have had induced abortions.
3. A similar example is given by Elard Koch. The method generated an estimated 165,000 abortions in Mexico City annually before legalisation in in Mexico City in 2007. But as Koch points out, in the year following legalisation, only 10,137 abortions were recorded in Mexico City, and the total number of abortions over a five year period following legalisation was 78,544, less than half the AICM estimate for one year alone – and this is after a likely significant increase as a result of legalisation. Either the method is wildly inaccurate, or the overwhelming majority of abortions remain illegal and potentially unsafe even after abortion is legalised (or both). Neither are promising options for those advocating abortion legalisation.
4. It is perhaps unsurprising in light of this that the Guttmacher Institute, from which this method mainly derives, admit: “If our assumptions about the likelihood that women seek such care [for miscarriage] are inaccurate, our abortion estimates will be as well”. The above evidence demonstrates conclusively that these assumptions are indeed incorrect.

The Royal College of Obstetricians and Gynaecologists themselves noted in 1966 that high estimates of illegal abortion in the UK prior to legalisation were wildly inflated: ‘It has been repeatedly stated that as many as 100,000 criminal abortions are induced in this country each year, and a more recent estimate is 250,000. These, and an earlier figure of 50,000, are without any secure factual foundation of which we are aware.’ They even noted explicitly that the majority of complications they saw (at least 80%) were, in fact, from miscarriages, not abortions. By contrast, a more sober estimate in 1964 suggested around 10,000 illegal abortions a year. England and Wales now have over 200,000 abortions a year.

Does welfare reduce abortion?

It is easy to see why one might think more welfare would reduce the abortion rate: a significant number of women cite financial concerns as a reason for abortion. So, alleviating those financial concerns might help them to continue the pregnancy.

Unfortunately, the empirical evidence on welfare is far more mixed, demonstrating minimal impact or occasionally conflicting results. One explanation of conflicting results is that welfare has different effects in different contexts. This appears to be true in the case of abortion.

It appears that welfare modulates and amplifies a jurisdiction’s general position towards abortion, at least in the much-studied US context. That is to say, in pro-life states which discourage abortion, welfare is indeed associated with reduced abortion rates. But in pro-choice states with minimal limitations on abortion, welfare is associated with increased abortion rates, meaning that pro-choice pro-welfare parties are not only worse than pro-life parties, but (with respect to abortion rates) even worse than a pro-choice anti-welfare party. The point is: without a pro-life cultural and policy context, welfare appears to be not only ineffective, but even counterproductive.

Evidence from outside the US is consistent with this picture. European countries with the most maternal leave and most generous overall family/gender policies have abortion rates significantly higher than the European average, especially the Nordic countries. And while abortion rates in the developing world are highly controversial/unreliable, it is uncontroversial that a far lower proportion of unwanted pregnancies in low income countries result in abortion compared with high income countries.

There are a variety of explanations for this phenomenon. For example, finances are a relatively uncommon reason for abortion, especially in low income countries. Second, most women who cite financial concerns also cite other reasons for abortion, so that alleviating these concerns will not necessarily make the pregnancy acceptable. Third, only a tiny proportion of women consider financial support relevant to their decision, and most say that nothing could have changed their decision. In the US, only a small minority of women said that they would make a different decision given a European-style safety net. These and other reasons explain why the empirical evidence for welfare improving abortion rates is very limited.

Does sex education reduce abortion?

Perhaps by definition, good sex education reduces abortion – since part of what makes sex education good is that it results in fewer pregnancies likely to end in abortion. The question most people have in mind for political reasons is: does the sex education typically offered in public schools reduce abortion rates? Again, the rationale is intuitive, but the empirical evidence is virtually non-existent.

Systematic reviews of randomised control trials from the prestigious Cochrane database have found no impact of sex education interventions on teen unintended pregnancy rates. In fact, the only randomised controlled trial which did show a significant reduction in teen pregnancy – with a massive effect size of 80% – was from a much-derided abstinence programme. More recent cross-country evidence has found an increase in teen pregnancy and abortion rates from sex education mandates, mitigated by allowing parents to opt their children out.

Again, this seems counterintuitive to many people. As Paton, Bullivant and Soto point out, however, sex education can have mixed effects: it can reduce the likelihood of a given instance of sexual intercourse resulting in pregnancy (through encouraging contraception), but can also increase the rate of ‘uncommitted’ sexual intercourse more generally (through removing fears of pregnancy, normalising extramarital sex, etc.). Hence, the effect of sex education is inherently mixed and can only be discerned by engaging the actual empirical evidence. When that is done, there appears to be no convincing evidence that typical sex education works – though some limited evidence that abstinence education works.

Does contraception reduce abortion?

One thing is obvious, so no one may interpret me as saying otherwise: contraception clearly reduces the risk of pregnancy in a given instance of sexual intercourse.

The relevant question, therefore, is whether contraception promotion a) leads to fewer unintended pregnancies, and b) affects the likelihood of an unintended pregnancy being aborted. These are separate questions. Regarding the latter, contraception may increase the likelihood of an unintended pregnancy being aborted, since it cultivates a culture where pregnancy is seen as only an optional outcome of sex (in other words, it makes pregnancies more likely to be unwanted, as explained by John Cleland).

Regarding the former, the same considerations apply as in the case of sex education: namely, contraception can decrease the cost of risky sex and thereby incentivise it. This may compensate for or even outweigh any decrease in pregnancies from using contraception in a given instance. In fact, there is clear empirical evidence that contraception increases risky sexual behaviour. Since contraception failure rates are significantly higher than most people realise (leading to half or more of all abortions in many countries), it may well be that the increase in risky sex outweighs the decrease in pregnancy-per-intercourse.

It is known that abortion rates are still very high even where contraceptive access is close to perfect. Indeed, Northern Europe, which has extremely good access to contraception (and has for many years), has particularly high abortion rates, indeed, some of the highest in the world.

Examining the empirical evidence in detail shows no consistent pattern between contraceptive availability and abortion. No study on emergency contraception to date has shown a reduction in abortion rates. A recent systematic review found no effect of contraceptive-promoting interventions on reducing adolescent pregnancy. Abortion rates increase with increased contraception just as often as they decrease, and those countries which do see an inverse correlation tend to be exceptional cases where abortion is already extremely widespread as a means of contraception, e.g. the former Soviet countries. Even where there is an inverse correlation, the effect size is still relatively small. Many studies looking at particular contraceptive access programmes show no effect on abortions or teen pregnancies, in some cases even showing a long-term increase.

Finally, it is worth noting that the contraception whose widespread use is most likely to cause a decrease in abortions – namely, the more effective long-acting reversible contraception – may itself be abortifacient at times, which would be the reason it is so effective in the first place. So for a pro-lifer the use of such contraception would be questionable (though the empirical evidence is still somewhat unclear).

It is clear that contraception promotion as a policy does not reliably reduce abortion rates because of these offsetting factors. But even if it did, it is not clear that this should be particularly salient politically. In fact, when the US President changes, there is in general no change in contraception use either in the US or overseas (see supplementary material, figure 4). Even in developing countries, unmet need for contraception is only 12.8% (it is 5.9% in the US). This is a small figure, but some might think that contraception availability might address that 12.8% and significantly reduce the abortion rate as a result. However, it turns out that only 4-8% of that 12% – that is, 0.48-0.96% of sexually active women of reproductive age – gave ‘lack of access’ as a reason for not using contraception. Generally, it was the choice of the women not to use it. Hence it is very difficult to believe that increasing access to contraception would have a very large beneficial effect on the abortion rate. Contraception access is relatively politically insignificant in comparison to pro-life laws.

Democrat Presidents reduce abortion rates more than Republican Presidents

As a piece of evidence for the above objections – that welfare, sex education and contraception reduce the abortion rate more than legal protections for unborn children – it is sometimes claimed that abortion rates fall quicker under Democrat Presidents than under Republicans. I write this response not out of any particular fondness for the Republican Party, with which I have deep disagreements on economic and environmental issues, among other things. But insofar as this claim is used to confirm the idea that welfare, sex education and contraception promotion work better than legal protections for unborn children, it is worth responding to.

Graphs like the following are normally shown:

The claim is that abortions increased under Ford (R) and Carter (D), plateaued under Reagan (R) and Bush Sr (R), decreased under Clinton (D), decreased more slowly under Bush Jr (R), and decreased more quickly again under Obama (D).

The biggest problem with such a claim is that the federal government, and even more so the President, have only a very small, or even non-existent, short-term impact on abortion rates, since Roe v Wade forces states to allow abortion, and the Senate filibuster (requiring a supermajority) means no significant pro-life legislation can pass at a federal level. And of course, for many of these years, the President had not even a Senate majority, let alone a supermajority. Ford had a Senate minority for all 4 years (and Nixon likewise before him), Reagan for 2 out of 8 years, Bush Sr for all 4 years, Clinton for 6 out of 8 years, Bush Jr for 4 out of 8 years, and Obama for 2 out of 8 years. Given that virtually all domestic pro-life or pro-choice legislation requires a Senate supermajority, along with Roe v Wade, the President has very little power to do anything domestically. The main things they can do are a) support or discourage abortion overseas, and b) appoint Supreme Court Justices. Obviously the former will not show up in domestic abortion rates, and the latter will have only a very long-term impact, given how long it takes to fundamentally change the composition of the Court. Hence, for example, it is likely (at the time of writing; January 2022) that Roe v Wade will be overturned under President Biden, a Democrat. This is obviously despite a Democrat President, not because of him.

This also holds a clue to the large drop in abortions under President Clinton. As can be seen from the graph, the large majority of the drop came at the very beginning of the Presidency (1993), with the rate then decreasing at a fairly average rate. But this was just after the 1992 Supreme Court judgment Planned Parenthood v Casey which, while upholding the legal right to abortion, allowed unprecedented regulations on abortion, such as informed consent laws, waiting periods, parental involvement laws, and reporting requirements, at a state level. Studies have shown repeatedly that these sorts of state laws reduce abortion rates. Again, clearly these laws were allowed to be implemented despite a Democrat President, not because of him.

This is an example of more fundamental problems with the original claim. There are simply too many other factors – most of which are irrelevant to the President – to make this sort of crude association. Under Ford, abortion had just been legalised nationwide by the Supreme Court, so this is clearly responsible for the big uptick in abortions, which continued under Carter. Reagan and Bush Sr appear, if anything, to have flattened out the sharp rise, and so can hardly be faulted for not reducing it as well, when the natural trend was sharply upwards. Clinton coincided with Planned Parenthood v Casey, and also implemented welfare reform which made welfare law much stricter, so the reduction in abortions can hardly be attributed to more welfare under a Democrat President! In some presentations of this data there are even more obvious problems: some graphs (depending on the source) show a big drop in the 1990s under Clinton which is simply due to the fact that California stopped reporting its 200-300,000 abortions per year in the late 1990s.

It does appear to be true that the rate fell faster under Obama than under Bush, but this is a very small sample size with Presidents who could do very little about abortion. Indeed, the start of Obama’s term came shortly after the Supreme Court upheld the partial birth abortion ban (2007), which coincided with a spike in pro-life public opinion. It is difficult to draw any conclusions either way here.

Given that the overwhelming majority of abortion legislation comes at a state level, what is far more telling is the state abortion rate. Below is a graph I put together with data from the Guttmacher Institute’s Data Center (the GI being the research arm of Planned Parenthood). States are colour coded by state legislature: red for Republican, blue for Democrat and yellow for divided. Three things can easily be seen: a) whether a state is red or blue does not affect its contraceptive prevalence; b) contraceptive prevalence seems to have minimal overall correlation with abortion rate; and c) there is a very clear difference in abortion rate between red and blue states – and the differences are enormous, with Republican states typically having abortion rates between 5-10 with a few higher, and Democrat states being fairly evenly distributed between 10-30 (and none below 10). Since states have far more power over abortion policy than the federal government, and even more so than the President, this seems to be a far better measure, and if anything would clearly indicate that pro-life laws are far more effective than the proposed alternatives to reduce abortion rates.

This is confirmed by the fact that European countries which have had very strong welfare systems, free contraception for many years, and widespread sex education in schools, such as France, the UK, and Sweden, still have significantly higher abortion rates than the US.

Of course, even this does not take into account confounding factors, though the correlation is impressive. Most telling, of course, is the rigorous studies we have simply showing clearly that legal protections for unborn children do, in fact, reduce the abortion rate, while welfare only does so in the context of legal protections, sex education does so only when it encourages delaying sexual debut, and contraception promotion does so (generally) only when abortion is already widely used as birth control (and perhaps a few other exceptional circumstances). See the questions on these above.

Finally, it is worth noting that the other main contribution Presidents can make is through foreign and aid policy: encouraging other countries to legalise abortion, or donating money to organisations which lobby for, or even provide, abortions in developing countries. Given the evidence seen above, it is clear that other countries legalising abortion will drastically increase the abortion rate in them, as has been seen in almost every country which has ever legalised abortion. It is pretty uncontroversial that Democrat Presidents encourage other countries to legalise abortion, and Republican Presidents encourage them not to. Likewise with money donated to abortion lobbying organisations such as Planned Parenthood and Marie Stopes.

There has been one challenge to this claim, however: it is claimed that a 2019 paper by Brooks et al. demonstrates that the Mexico City Policy (which bans US aid funding going to any organisation which performs or promotes abortion overseas) has a counterproductive effect. It is alleged that by reducing contraceptive prevalence (by defunding family planning organisations), the abortion rate is paradoxically increased. There are a number of problems with attempts to draw these conclusions from this paper.

1. The paper doesn’t take into account changes to the law, because these are far rarer and more discrete. This has an enormous impact on abortion prevalence, and is far more likely to occur under American pressure under Democrat governments.
2. Only a tiny minority of women in developing countries lacked access to contraception by 2008, the end of Bush’s tenure. As explained in ‘Does contraception reduce abortion?’, only around 0.48-0.96% of women in developing countries were not using contraception due to lack of access in that year. Given that some women will also lack access to abortion under Democrat governments, it is profoundly implausible that this small change in access could increase abortion rates by the 40% alleged by the authors.
3. Contraceptive prevalence is, in fact, not affected at all by the single Republican government examined in the study. The figure shown is taken directly from the study, and shows countries strongly affected by the Mexico City policy in turquoise and countries not strongly affected in orange. In both sets of countries, contraceptive prevalence continued increasing under Bush Jr, at about the same rate as it was increasing under Clinton previously. In fact, the only different trend observable is that in the middle of the Bush administration, contraceptive prevalence rapidly accelerated far beyond the rate of increase under Clinton. This increase plateaued again under Obama. Indeed, though countries affected by the MCP had significantly lower contraceptive prevalence at the start of Bush’s tenure than countries not affected by it, by the end of Bush’s tenure countries affected by the MCP had almost caught up. Hence, if anything, the study seems to show that contraceptive prevalence increased at a greater rate during the implementation of the Mexico City Policy.

4. It is independently known that funding restrictions have a huge impact on lowering abortion rates. The Guttmacher Institute’s own literature review shows that cutting Medicaid funding for abortion prevents a massive 18-37% of abortions, with the best study suggesting 37%.

5. By contrast, it is independently known (see ‘Does contraception reduce abortion?’) that contraception promotion has at best a mixed effect on abortion rates.

6. In countries not affected by the MCP, the abortion rate was very high but quickly decreasing under Clinton and Obama, and bottomed out under Bush in between (see picture). In countries affected by the MCP, abortions were steadily increasing under Clinton, plateaued under Bush, before dramatically increasing again under Obama. The clear trend is that in both sets of countries, abortion rates were significantly lower under Bush (except insofar as a trend of slight increased continued under Bush from Clinton in MCP countries). Obama’s presidency corresponded with a dramatic and fairly immediate spike in the number of abortions, despite contraceptive prevalence increasing during that time.

7. These fairly clear trends are obscured by the methodology employed the authors, which used the non-MCP countries (orange line) as a control, such that any deviation from the same trend in the MCP countries (turquoise) is interpreted as resulting from the imposition (or not) of the Mexico City Policy. But this methodology is only appropriate when the control and intervention groups are relevantly similar and can be expected to follow the same trends. In short, the difference-in-difference methodology assumes that the trends in both groups will be exactly parallel apart from the intervention studied. But the graph itself shows that this is an impossible assumption. Non-MCP countries started with an abortion rate around three times the rate of MCP countries. Non-MCP countries were rapidly decreasing, so quickly in fact that for MCP countries to follow the same trend, MCP countries would end up with a negative abortion rate – which is obviously impossible. The two groups are completely different and had completely different natural trajectories: it follows that the methodology employed by the authors is wholly inappropriate and should not have been used.

For all these reasons, the claim that the Mexico City Policy increases the abortion rate by reducing contraceptive prevalence has no reasonable empirical basis at all.

In summary, the available evidence suggests that, in fact, Republican governments are far more likely to reduce abortion rates – as reflected most clearly in state abortion rates – than Democrat governments. This is not to offer any comment on the parties more generally. Rather, it is to dispel the notion that legal protections for the unborn are less important for reducing abortion than other measures such as contraception, sex education and welfare.