More Pregnancy, Less Crime

When it comes to crime, economists focus on deterrence. Deterrence works but it’s not the only thing that works. Simple things like better street lighting can reduce crime as can high-quality early education or psychological interventions such as cognitive behavioral therapy. The sociological literature has emphasized that crime is about preferences as well as constraints. Life-events or turning points such as marriage and childbirth, for example, can greatly change crime preferences. The sociological literature is mostly from case studies but in an excellent new paper, Family Formation and Crime, Maxim Massenkoff and Evan Rose (both on the job market from Berkeley) demonstrate these insights in a huge dataset.

A big part of what makes their paper compelling is that almost all of the results are blindingly clear in the raw data or using simple analysis. Here, for example, is the crime rate for women (drug, DUI, economic, or property destruction crimes) in the years before pregnancy, during pregnancy (between the red dotted lines) and after birth. Crime rates fall dramatically with pregnancy and in the three years after birth they are 50% lower on average than in the years before pregnancy.

Pregnancy imposes some physical limits on women but the effects are also very large for men whose crime rates fall by 25-30% during pregnancy of their partner and continue at that lower rate for years afterwards. Keep in mind that in our paper on three strikes, Helland and I found that the prospect of an additional twenty years to life (!) reduce criminal recidivism by just ~17%, so the effect of pregnancy is astoundingly large.

It’s not obvious what the policy implications are. Have children at a younger age doesn’t sound quite right, although in an analysis on teen births Massenkoff and Rose do indeed show that whatever the costs of teen pregnancy there are some offsetting benefits in reduced crime of the parents. More generally, however, there are policy implication if we think beyond the immediate results. First, these results show that crime isn’t simply a product of family background, poverty and neglect. Crime is a choice.

In Doing the Best I Can: Fatherhood in the Inner City, Edin and Nelson relay the following anecdote (quoted in Massenkoff and Rose):

Upon hearing the news that the woman they are “with” is expecting, men such as Byron are suddenly transformed. This part-time cab driver and sometime weed dealer almost immediately secured a city job in the sanitation department (p. 36).

Byron chose to change and he did so based on the rational expectation of a future event. Massnekoff’s and Rose show that these choices are common.

Instead of thinking of these results as being about pregnancy and marriage we should ask what is it about pregnancy and marriage that makes people reduce crime? Love, responsibility and long-run thinking are all at play. In economic terms, pregnancy reduces discount rates and gives men and women a reason to invest in human capital and work for the future. Children and marriage play a large role in socializing and “civilizing” both men and women but they surely can’t be the only such factors. Indeed, although men and women on average reduce their crime rates dramatically on pregnancy this is mostly coming from men and women who had high rates to begin with–there are plenty of men and women who don’t much reduce their crime rates on pregnancy because they were already low–in a way, these men and women were pre-socialized so how do we extend the benefits of pregnancy to the expectation of pregnancy or how can we widen the effect to other factors that can also civilize?


So we should expect that men who produce lots of children at a young age, with multiple women, would be a cohort not associated with a high crime rate? Whereas people who delay pregnancy, say until after grad school, are at high risk?

Obviously not.

It’s saying for group of people A who commit crimes at a rate of Y / week, pregnancy will lower that crime rate to Y-X / week.

Group A includes the sperm donor.

That's not correct. The groups of people identified in each chart are "first-time mothers" and "first-time fathers."


Your missing the point that the left often misses that what causes criminality also causes poverty. So among the crime prone group want things reduce crime, among the answers is parenthood and marriage.

Exacty. And as it was noted below, the trick here is "being invested". If a guy fathers multiple kids with multiple women his investment will very likely be low. That is why marriage "works", by the way.

I think the father has to be invested in the child for that trick to work. A man who has lots of children but is not expected to take care of them has no reason to change his time preference.

George Gilder.

>When it comes to crime, economists focus on deterrence.

Well, that's not true. They focus on how quickly we can let people out of prison.

Not if they can do useful work whilst incarcerated for the private company running the prison!

Who needs slavery when you have prisoners-with-jobs?

It makes sense intuitive sense that pregnancy would decrease crime rates for women, however Steven Levitt demonstrated with significant statistical research that the age and financial security of a women during pregnancy is correlated to the likelihood of the child committing crimes in the future. So the offset of having many more young pregnant women may be a time lag in increasing crime rates.

Yeah... encouraging criminals and near-criminals to have kids seems like a disastrous long-term policy.

'Significant statistical research'

Heh heh

Correlation ain't causation, Sidney!

Levitt guessed. That is all.

So we should expect that men who produce lots of children at a young age, with multiple women, would be a cohort not associated with a high crime rate--like blacks, muslims and black muslims?

And we should expect that men who produce no children, like homosexuals, would be a cohort associated with a very high crime rate?

I would expect the same results among poor whites. Don't pull my comment into your racism.

Why would you expect that? Poor whites don't commit much crime.

West Virginia is one of the poorest, whitest, and most peaceful states in the US.

IIRC Haitians (and rural blacks?) don't commit much crime either.

Actually, West Virginia used to be one of the most peaceful but now it is roughly in the middle. Since the early 1990s, violent crime has trended down in the U.S. but has risen steadily in West Virginia.

Thank you for pointing that out. I just googled all these facts and figures and it is like you are saying. That is very interesting.

Homosexuals do not, as far as I can tell, have higher rates of crimes, certainly not higher rates of poverty.

Though I would caution anyone here from assuming gays have *no kids.* It's long been my position that gays have fewer kids. Though a mathematician from Harvard recently challenged that contention. But gays DO have kids through heterosexual coitus, artificial insemination and adoption.

Muslims do not have a high crime rate.

Who knew? Well, everybody, for generations.

The policy implications are actually pretty obvious.

Yes: stop punishing the poor for wanting to have children. Provide support for young families so children aren't born into poverty. Take the money currently invested in the war on drugs and the prison system and use it to support young families, including unmarried men and women who have children together. Show young men that the path to social and economic stability lies in settling down and having kids.

Oh sorry, I forgot I was on a libertarian blog. More prisons! They are very good for liberty! Social programs for poor people are anti-liberty!

Putting people on the dole doesn’t seem to solve these problems. See for example Life At The Bottom, which discusses the experience in England.

All the social norms which encourage and sustain family formation, and have been and continue to be systematically dismantled by progressives, do seem to work.

Thank the government for its misguided policies on welfare and child support for the demise of the low-income family. A sexually active female knows that her baby is source of income, especially if she can chase the father off. No need to listen to his crap or pick up after the slob. Not only that but the benevolent state puts its hooks into his financials and ruins his life for at least 18 years because her congress with the dope doesn't involve any commitment on her part. If the AFDC was destroyed by an asteroid there would be more and better families.

By saying, "A sexually active female," it's fairly evident that you don't know much about women or anything else.

I would like to apologize for the libertarians of this blog. For far too long we have been unwilling to violently expel our neighbors from their homes so needy families can occupy them.

For that, as well as our many other perceived faults, I truly apologize.

I’m curious. I don’t really like my neighbor as he is kind of unfriendly. But his wife is nice and his little girl is just adorable. What kind of terrorizing would you suggest so I can get them to behave in a way I want? I’ve tried mowing his lawn when they were in the hospital giving birth, and I plowed their driveway multiple times, but I still can’t get them to take in a needy family. How were you able to convince your neighbors?

Government social programs for the poor are anti-liberty and rarely work. Libertarians and (neo-) conservatives have been saying and showing this for decades.

It's also social. The male is interacting more with the wife than with other males. In other words, you are not engaged in social activities with other males in gangs. You are at home watching TV and changing a baby's diaper.

If there were an activity for males that did not involve interaction with delinquent males or competition for supremacy by engaging in risky activity or illegal activity, that might work as well.

US Army.

The vast majority would never make the cutoffs to enlist.

It’s not 1940. Nor, for that matter, 1967.

Evidence please as to "vast majority".

I’d break it into the three main buckets.

1) high school diploma
2) IQ threshold
3) no previous criminal history

1) Only 16% of prisoners have a high school diploma. So that requirement slices out a good chunk. The criminal rate for those who graduated high school is already much lower.

2) Minimum IQ is 92, but for what they’re actually recruiting for it’s closer to 100. Chronic offenders are closer to low 80s. Most wouldn’t meet this threshold.

3) Lack of criminal record. Standards dropped in 2005 with waivers. That’s already been cut. So we’re assuming these people were committing crimes and then not caught? Seems unlikely.

Throw in ability to pass a drug test and physical fitness requirements and I’d say yes, vast majority.

I search for you 16% but found this:
The results show that 30 percent of incarcerated adults had attained less than a high school diploma—twice the percentage for U.S. households (14 percent). And more incarcerated adults scored at the lowest levels in both the literacy and numeracy assessments (see chart).

Skeptic, What you did was limit your population to the subset of incarcerated individuals and went from there.

The post was not about incarcerated individuals.

And, by the way, Skeptic, a criminal record does not automatically exclude from the military:

"Nearly 12 percent of U.S. Army recruits who entered basic training this year needed a special waiver for those with criminal records, a dramatic increase over last year and more than twice the percentage four years ago, according to new Army statistics obtained by the Globe."


Your data is from 2007, at the height of the Iraq war. Quoting myself here:

Standards dropped in 2005 with waivers. That’s already been cut.

That was a temporary lowering of a standards to fill the meat grinder. Those standards have been brought back to normal.

From more recent studies:

"71 percent of young Americans between 17 and 24 are ineligible to serve in the military—that is 24 million of the 34 million people of that age group."

So, everybody must be committing crimes or incarcerated, have low IQ or no high school diploma.

I’ll quote myself again:

Throw in ability to pass a drug test and physical fitness requirements and I’d say yes, vast majority.

But, then, that statement would be true of the entire 17-24 population, and not the subset of the post addressed.

Really weird comments that are unsupported or, now, tautological, given the correct assumptions.

Not if the variables are highly correlated? Low IQ, criminality, and BMI are closely clustered together.

Might bring you from 71% to closer to 95%. Aka vast majority.

I guess 71% of the 17-24 year olds are a vast majority as well. Let's face it, you do not want to accept that social interaction with other males who engage in criminal activity has no effect, but, rather, it is just the fact that 71% of males do not qualify for the military leads you to conclude that that is the explanation. AKA tautology.

I think for 1) that’s a fair criticism. I don’t have the time to dig further into crime rates for high school diploma vs dropout. I think the point still generally stands. The ones committing the crime will be much less likely to have a diploma.

2) and 3) are still completely valid.

We can’t know for sure unless they break down crime rates further by IQ and diploma status.

But yes fair point on 1).

Computer games?

the red dotted lines indicate the pregnant woman has become an insecure collateralized debt option, as such with or without a high school diploma, the lines indicated intransience in the social ladder. Which is what Alex is trying to say.

Unless crime rates fall for women who get abortions, the title should be "More Mothers, Less Crime".

'Simple things like better street lighting'

Most people consider this deterrence, actually, since it makes the likelihood of getting caught and punished much higher.

Turning points such as marriage and childbirth can greatly change pretty much the life of everyone who experiences them, but then, such banal observations always sound boring outside of a research paper.

They don’t sound boring, they sound like unsubstantiated opinions.

I am guessing you have neither been married, nor raised a child.

And completely ignored all of the presented data.

Well your unsubstantiated educated guess would be wrong. Been married for more than half of my life and had children prior to my being able to vote even. In fact, I actually agree much with your opinion... as my first child was very much a turning point in my life. My point, while inappropriately snarky (my apologies for that) was that anecdotes and educated guesses are not structured investigations applying the scientific method.

So everyone needs to go look at the graphs at the end of the paper. The drop in crime for men after first birth is not as dramatic as the event study coefficients would suggest and is most likely just due to men getting older. The effect goes away for second births. And marriage seems to negate the effect of pregnancy and birth for both mothers and fathers (i.e. marriage is bad for crime). Also, birth is really bad for domestic violence (as a parent, I can see why). So is marriage (evidence for adverse selection in mating markets?).

Here's what I would have liked to have seen. Instead of doing the event study analysis and raw averages of offenses, show percent of offenders for each crime category that are mothers or fathers, or you could provide a density plot for where the x-axes is months from birth. Then provide a logistic model for the probability of arrest with time from birth as the coefficient of interest.

Also, they use a 4th order polynomial for age and the dummies "age 18" and "age 21". Why?

I believe there are studies that showed that men who became fathers experience a testosterone drop, and if I recall, those with the highest levels before having a child had the steepest drop afterwards. The more time spent involved with children, the lower the T levels. Certainly that might have an impact as well with regard to aggressiveness and impulsiveness, and thus criminality.

I find it very hard to imagine how "quality early education" would effect crime rates 8 to 10 years later, I want much more evidence.

As Geralt commented, new fathers (especially involved ones) experience a permanent drop in testosterone. That must be driving much of these results.

New mothers (especially ones who have an unmedicated labor and birth) experience even greater hormonal changes. Prolactin is the "mothering hormone" that causes nurturing behavior and milk production. Oxytocin is the "love hormone" that causes bonding and uterine contractions. From personal experience, pregnancy and birth changed me in ways that merely "having" a baby in my lifestyle could not account for. These hormones were totally transformative.

The significance of these biological factors does not give much hope for being able to replicate the results without actual pregnancy and birth. For example, exogenous oxytocin does not cross the blood-brain barrier, so although it is easily synthesized and administered (and often is), that only produces uterine contractions, not bonding - and in fact, adding oxytocin to the bloodstream decreases the brain's production of endogenous oxytocin that would cause bonding. So it's really hard to replicate nature.

Oxytocin boosts favoritism toward one's own group- tribalism. Being family-oriented is a beautiful thing, until it's not.

This is my first time seeing "loving your family is just an extremely specific kind of racism" in the wild.

This is why the Economics profession is broken. Why spend time and resources investigating useless knowledge? What is actionable here?

The use of public policies to promote pregnancy so we can reduce crime is on its face crazy, so why investigate this at all?

I would bet some "thinkers" around these parts are coarse enough to suppose that inducing pregnancy through the water supply would be a good way to reduce crime.

What happens to the crime rates if the male subsequently separates from the woman and child? What happens to the crime rate as the child grows beyond, say, 5 years old?

Free vasectomies: discuss!

In a magazine article I read many years ago, a young family in some Central American city was introduced. The husband had been part of some rebel group, was captured and convicted, then told by the judge that the prison sentence would be suspended if the young man would go get married. A shrewd application of domesticity to induce domestication, which in the article the couple seemed happy with.

Nice if they got a happy end, but... troubled guy has to marry to avoid prison looks both like an oddball comedy and the recipe dor a domestic disaster. As unconventional sentencing foes, shipping him to the army would be safer.

In this man's case, he had been fighting the army. Not standard criminality, but also fueled by the freedom of youth. The idea seemed to be to tie him down with a family, make him the opposite of a "free man."

Ok, Boomer. Yet, in the last two years, Brazil drastically lowered its homicides rate wirhout boosting its births rate.

Yes, and the US has drastically lowered its homicide rate the last few decades even as the birth rate has fallen (and births gotten lower.) That one would still get this result on the one particular difference even when the underlying variables are moving in a different relationship is remarkable.

Wonder if they dug into employment per and post pregnancy as well. I suspect there is still something of a bias for hiring people with families as they tend to be those who have a bit less freedom to just walk off the job. I suspect there are some other aspects that might relate to team work aspects and understanding responsibility that go with that/are signaled by that.

I haven't read the paper so this may be addressed but is pregnancy treated as a random event? Maybe pregnancy is simply a signal that leads to a separating equilibrium but is not causal? I wonder if there's a comparison between those who get pregnant and: carry to term, intentionally don't carry to term, or unintentionally don't carry to term.


Interesting comment. But... seems as though there could be a causal relationship between those that don’t carry to term on purpose and though. If one is willing to sacrifice their own offspring for economic convenience or to avoid personal responsibility for their actions... who wouldn’t they cheat given the opportunity. What greater crime is their than terminating the life of ones own innocent defenseless offspring? Those that would rob their own children of the right to live for money would certainly seem capable of many other lesser crimes undertaken in pursuit of dolla dolla bills y’all.

I think the observation Kevin makes is about causality. For example, having a baby (becoming impregnated) is probably associated with couple activity outside of engaging in activities with other males who may be engaged in gang or other, if you are spending more time with the female, you are less involved with male cohorts. So, that might be the explanation (less male male activity), and not pregnancy per se. What this post really misses is the behavioral econ and social network analysis that might form alternative hypotheses.

In other words, pregnancy may just be the end point and not the cause.

Mea culpa. That was vino logic there.

Yes, the women have little bastards who commit the crimes..... mind-bending... fruits of the welfare state..

The obvious conclusion from this paper is: force every male who is arrested for a crime to become pregnant and bear a child.

This is actionable. right?

More affordable family formation along with stronger marriage and weaker divorce norms and laws would help quite a bit.

How do those actions make males pregnant?

I suspect that some types of pregnancies (e.g., within wedlock) are better for society than other types of pregnancies (e.g., out of wedlock). Does the paper elucidate upon that?

Cool, now when do we get to talk about crime rates by race ?

Comments for this post are closed