Book summary: Selfish Reasons to Have More Kids

by Optimization Process6 min read7th Oct 202110 comments


World Modeling

Bryan Caplan's Selfish reasons to have more kids: why being a great parent is less work and more fun than you think was recommended to me as a strongly-science-supported take on childrearing. I found it valuable; here are my take-aways.

TLDR: Variation in parenting (within reason) has no significant long-term effects. So take it easy, and maybe invest some of that saved energy in another kid.

The main thrust of the book is:

Nature >> Nurture

I mean, Caplan says it with somewhat more caveats, but that's the bottom line from his chapters on behavioral genetics. He cites a big pile of twin studies and adoption studies, and tries to answer, for various attributes:

If you're 80th-percentile on the scale, what percentile should you expect your adopted sibling to be in? How about your long-lost identical twin?

The answers, summarized:

Attribute Adopted sibling's percentile Identical twin's percentile
Life expectancy ~50 58
Health (self-reported) ~50 56-65
Health (objective) ~50 56-65
Dental problems ~50 56-65
Height (6in above average) ~50 5in above average
Weight (20lb above average) ~50 15lb above average
IQ 51-55 70-80
Misc intelligence ~50 70+
Happiness ~50 67
Education 51-55 75
High school GPA ~50 71
Income 51-55 "about twice as similar as fraternal" (like you'd expect from pure genetics)
Conscientiousness, agreeableness ~55 "unusually large effects"
Political/religious behavior[1] 56 (not given)
Political/religious labels 73 (not given)
Filial appreciation ~60 (not given)

And a couple of attributes that don't fit neatly in to that table:

  • If a kid is born to parents without criminal records, then being adopted by parents with criminal records (vs adoptive parents with no records) raises their chance of getting a criminal record by 1.2% (from 13.5% to 14.7%).
  • Smoking/alcoholism/drug use have mixed results. One study shows a moderate-to-large nurture effects on alcohol use, one shows none.

Note that the world is really complicated, and each of these traits has several studies measuring slightly different things, and the whole thing is riddled with quirks and caveats that don't fit into this table. Does it make sense to separate out marijuana use from amphetamines/cocain/sedatives, or is that p-hacking? What about subgrouping adopted children based on whether their biological parents have criminal records? I dunno, man. Reality is often underpowered.

But also... look at all those places where upbringing fails to show an effect. The only nontrivial effects are on (a) how much appreciation they'll show you, and (b) their political/religious labels. And that's without significantly affecting actual political/religious behavior, meaning the labels are basically just lip service![2]

(Caveat: all of the above is about long-term effects, i.e. effects on the grown children. Environment can have short-term effects on a person's IQ, income, criminality, and sexual behavior; but it fades out almost completely by age 25, and, for most of those traits, much sooner.)


Anyway. If you accept the table above—pretty much everything you care about is basically nurture-independent—then what follows?

  1. Instead of aiming to be a 90th-percentile parent like everybody else, slack off! Be a 10th-percentile parent instead! Maybe even 5th! (Not a 0th, though: you want to stay within the domains of these studies, i.e. middle-class first-world families. Obviously, if you starve your kid, their body won't just learn to metabolize the amazing genes you gave them.)

    Take advantage of take-out meals and electronic babysitters! And real babysitters, or a nanny if you can afford one! Don't worry so much about the kid dying: kids are safer than they've ever been, and the chances of death from accidents or homicide are under one-in-a-million per week from years 1-24! (Year 0-1 is, admittedly, several times higher.) Are you stressed out from driving your kid to all their extracurriculars? Drop any they don't enjoy—in the short run you'll both be happier, and in the long run it doesn't matter anyway! Do they enjoy all their extracurriculars? Drop one anyway! The benefits from you being less stressed and irritable will very likely outweigh the fun of whatever you cut.

    Note that the advice to slack off comes from consideration of long-term effects and how, roughly speaking, nothing you do has any. Slacking off in ways that have short-term effects, ways that make your kid's childhood less happy is, obviously, costly, and you should only do it if it proportionately benefits you. Your kid's short-term happiness matters, and/but so does yours!

  2. Okay, so, parenting doesn't have significant long-term effects; but you know what does? Genes! Giving your kids good genes is worth a lot. And your most powerful tool for affecting your children's genes is choosing your mate. "Choose a spouse who resembles the kids you want to have," sayeth Caplan. "The right spouse is like a genie who grants wishes you are powerless to achieve through your own efforts."

    (Maybe this doesn't much change your optimization target—the traits I want in my children are pretty much the same traits I want in a mate, except for, uh, sex stuff—but it seems like a valuable frame to have consciously available.)

Everything else

The book's section on genetics was the part that really impressed me with its thick coating of detailed scientific study. Most of the rest of the book is a smattering of miscellaneous parenting advice, much like other parenting books I've read; but I think Caplan's opinions are more worth listening to than most folks', so I'll relay the particularly interesting bits.

  • Consider the long term. When you think about "having kids," you probably think of the first five years, or maybe the first twenty: those years of immense investment, of six-hour nights, of mental impairment on par with being legally drunk. To be fair, there's plenty of joy in there too, sure, but it's definitely a mixed bag.

    You probably don't think of the benefits you'll reap over the following thirty years, watching your kids go through life and playing with your grandchildren.

    Don't forget to weigh that too!

  • Life-giving science. There are a lot of technologies to help you have more/different kids! We've got:

    • in vitro fertilisation ($12k [edit: per cycle, needing on average 2-4 cycles])
    • artificial insemination ($400)
    • surrogacy ($100k [including medical expenses etc.], $20k more than if you were planning IVF anyway)
    • genetic screening (book doesn't say; a quick google suggests $5-10k)
    • receiving donated sperm ($400)
    • receiving a donated egg ($10k)
    • sperm sortation ($3000 to choose the gender with 75-90% accuracy)

(There are also chapters on how to get grandchildren and whether your kids are good for the world, but everything there seemed pretty common-sensical to me, so summarizing them would feel like a chore.)


About 60% of the book's value to me is captured in the nature-vs-nurture table up above, and the rest of this post captures another 20%. But the remaining 20%—the quirks and caveats, the citations, the section on grandchildren—might justify reading the actual book!

  1. I'm lumping politics and religion together because they follow exactly the same pattern in the studies Caplan cites. It's eerie, honestly, and... highly suggestive. ↩︎

  2. Political parties, at least in the U.S., seem very strongly opposed; how can you flip somebody's label without affecting their attitudes or behavior? Caplan's answer seems to be that, because most people are only mildly-to-moderately interested in politics, they can bat almost equally well for either team; and maybe highly-politically-active people break this pattern, but are rare enough that they don't change the statistics much. I don't know, I'm not fully satisfied with this, but I have no better explanation. ↩︎


10 comments, sorted by Highlighting new comments since Today at 11:33 PM
New Comment

Something I found very interesting/important in the book that you skipped over was the bit at the very beginning where children were asked what they wanted from their parents, and rather than asking for more extracurriculars or later bedtimes or anything like that they said that their parents seemed stressed/sad and they wanted their parents to be happier. I found that very lovely/compelling.

Hi - I like this post and I'm glad you were able to put 60% of the value of a book into a table! One question I had - you say that IVF costs $12k and surrogacy costs $100k, but also that surrogacy is only $20k more than IVF? That doesn't add up to me.

Also, sperm/egg donation are usually you getting paid to give those things, which help you have children technically. But those children are probably not being raised by you, so a lot of the benefits you cite, like playing with grandchildren, might be smaller for children created with donated gametes than children you bear and raise yourself.

I agree. I think the IVF number is just plain wrong. I'm getting ready to have IVF myself and the total bill will be well over $25k even if we succeed in the first round, which is only 65% likely.

Maybe he researched the cost of "IVF" itself, but didn't think to add on the cost of implantation, injectable drugs, etc. which is a huge percentage of the cost.

Ooh, you raise a good point, Caplan gives $12k as the per-cycle cost of IVF, which I failed to factor in. I will edit that in. Thank you for your data!

And you're right that medical expenses are part of the gap: the book says the "$100k" figure for surrogacy includes medical expenses (which you'd have to pay anyway) and "miscellaneous" (which... ???).

So, if we stick with the book's "$12k per cycle" figure, times an average of maybe 2 cycles, that gives $24k, which still leaves a $56k gap to be explained. Conceivably, medical expenses and "miscellaneous" could fill that gap? I'm sure you know better than I!

I'm saying it's $25k PER CYCLE. (granted, this is Bay Area prices, but still)

IVF requires multiple other expenses that aren't the fertilization itself. These other expenses include about $5-6k of injectable drugs that stimulate egg production, and about $6000 for the implantation.

you say that IVF costs $12k and surrogacy costs $100k, but also that surrogacy is only $20k more than IVF? That doesn't add up to me.

Ah, yes, this threw me too! I think @weft is right that (a) I wasn't accounting for multiple cycles of IVF being necessary, and (b) medical expenses etc. are part of the $100k surrogacy figure.

sperm/egg donation are usually you getting paid to give those things

Thanks for revealing that I wrote this ambiguously! The figures in the book are for receiving donated eggs/sperm. (Get inseminated for $355, get an egg implanted in you for $10k.)

Variation in parenting (within reason) has no significant long-term effects.

This runs contrary to everything I've read about attachment theory and also the main reasons given to oppose corporal punishment of children. Attachment theory states that the first years are essential where the child learns an attachment style based on the way they interact with their caregivers, that will largely influence the way they bond when they are adults. Alice Miller, for example, has been one to oppose child abuse (defined very broadly) partly on the grounds that it causes long-lasting psychological damage to the future adult.

This reminds me, one thing that's clearly proven is that young children have especially good abilities to learn new languages. As they grow older, their hearing tunes to the languages they hear, so that they become less and less able to distinguish subtle sounds in other languages.

So basically what Bryan Caplan is claiming is that all of the theories that say there are huge effects visible in people over 25 years old from early childhood parenting (at least from differences in early childhood parenting that are part of the normal distribution in our society) are wrong. Caplan is saying they are wrong because identical twin adoption studies show that identical twins raised apart are very similar on the measured variables, and also that the measured variables capture what we actually care about.

Does this actually contradict the evidence base saying that, having been abused as a child, for example, is bad for adults? 

First I don't know the literature on the effects of child abuse in adults. It is clear that correlational studies are not going to be able to tell you whether bad outcomes are due to some other factor (genetics, class status, local culture) that is strongly associated with also being abused as a child. But perhaps some of these studies use identical twins, or have found another reliable way to disentangle causality and reliably show that it was specifically the child abuse that caused the later problems. 

Assume that child abuse does cause 'long-lasting psychological damage to the future adult.' Or alternatively lets simply look at the very well attested point that children are better at learning languages than adults. Do either of these outcomes disagree with Bryan Caplan's argument that what you do as a parent doesn't matter very much?

I'd argue that Caplan's actual argument is still solid. The reason that it is still solid is that you are referring to long term effects from extreme changes in how the child is treated, rather than from switching from a high stress/ high pressure version of normal middle class parenting to a low stress version of normal middle class parenting. Ie from lots of extra curriculars to a few extra cirriculars. 

The evidence base Caplan depends on with identical twin studies by construction only involves parents who were able to convince an adoption board that they would be responsible. There are unlikely to be very many extremely bad or weird parents in this group. The evidence for the result only holds within this range. And Caplan's claim also was only that you can aim for the low stress part of this range because being a very good parent inside this range doesn't make much difference relative to being a below average parent within this range.

I find it misleading that what Caplan calls "variations in parenting" would actually mean "switching from a high stress/ high pressure version of normal middle class parenting to a low stress version of normal middle class parenting". You oppose that very limited definition on the one hand, and "extreme changes in how the child is treated" on the other hand, but there are many third options. Homeschooling, for example. Or raising a child in the countryside as opposed to a crowded city. I have nothing against the limited version of Caplan's argument, but I'm concerned readers might be misled into updating towards a bigger version of the argument.

Some questions regarding Personality: If IQ (often a sub-in for Openness), Conscientiousness and Agreeableness are often more likely to be closer to genetics, what about Extraversion and Emotional Stability? Those two effect organizational efficiency (similar to Agreeableness).

Also, the nature vs nurture debate table did not include anything on sexual selection (age of sexual debut, sexual activity, mating vs parenting effort), which can affect multi-generational planning. Some postulated that fraternal birth order and birth control use can mutate one's personality, leading to more varied results.