Seems plausible enough, but must be a quite small fraction of TFR decline due to this. It's kinda the worst case scenario. Maybe there are 50 different small causes of TFR decline (X different policies, Y different technologies, Z different changes in culture etc.), and we will never be able to estimate them using the noisy data we have.
> No article I’ve seen did a systematic comparison of different proposed variables, so it is somewhat difficult to judge the relative evidence for them.
The authors (perhaps in this paper, perhaps in discussion elsewhere) did a substantial review, precisely so as to isolate some variable that _could_ be judged accurately at all, in which the null hypothesis of mere noise did not dominate. Thus, you may be interested in their process/results.
Religiousness is a strong correlate at the individual level, but country level data for religiousness is sparse/spotty. By request from another person, I tried out the dataset I could find with the most coverage. It does show small positive effects of religiousness, but given the small amount of data, it's not particularly convincing and the effect size is small. https://x.com/KirkegaardEmil/status/2043375308204568624
Seems to me it's a version of Baumol's cost effect. As the world gets richer in absolute terms, both the absolute cost (in education cost, child care, etc.) opportunity cost (of mom's lost earnings, career setback, etc.) of having a kid is higher than ever (as there are no productivity gains in child birthing or child rearing).
Another way to think about it - income vs wealth effect. We are wealthier than ever (which would predict higher consumption of child-having), but having a kid reduces your income so much relative to those who don't have kids (again both in absolute and relative terms) so much that it swamps the wealth effect.
You could also think of it the same way as with the obesity. People are eating too much, in part because there is just so much more diverse, tasty and cheap food around, delivered to your door using only your phone. It has never been easier to eat in other words, so people eat more. Similarly, life offers much entertainment and many things to try. It has never been easier to try new things, travel practically anywhere, live there, 1000s of obscure hobbies are showcased on the internet, and endless free entertainment. So relative to these life offerings, spending 9 months being pregnant and dealing with infants/toddlers for years doesn't seem so relatively appealing as it once did.
Why is Israeli TFR so high even among more secular jews? My guess is ultra-nationalism, ethnic paranoia, high overall religiosity and pro natalist policies, things any state could in theory emulate but are not ideologically kosher in any Western state.
Ukraine is in a far more exists risk situation then Israel and it hasn’t helped TFR. South Koreas response to the Korean War was to try and suppress its fertility.
Maybe part of the answer is that Israel provides a lot of financial support to families, has a good school system, and free IVF. They also have a low divorce rate for a modern western country (they don’t have no fault).
War and terror plays a role. I would guess, 5 years after Oct 7, 2023, fertility will be higher than just before Oct 7, 2023. Because there is fear that children (or later, adults) may die in a war.
There is no butchery in Gaza. It's all islamist propaganda. Ukraine is interesting though. Maybe because it is for a lot of Ukrainians a war of choice that they haven't chosen.
I have been looking on the first plot (log vs normal) and I would have agreed with everything you said about it if the axis were switched!
For OECD countries it does not make difference either way. But the current plot suggests that for world for certain models the log variant has higher error (RMSE) than the non-log variant.
It even said so in the legend (below the line the log models are better).
Is the issue that you flipped the axis in the plot or you switched log and non-log in text? I guess the latter based on the intuitive reasoning, but I am not really sure.
https://www.journals.uchicago.edu/doi/abs/10.1086/731812?journalCode=jle
(This paper is primarily about isolating a robustly causal variable, not whether this variable is the most consequential in TFR decline.)
Seems plausible enough, but must be a quite small fraction of TFR decline due to this. It's kinda the worst case scenario. Maybe there are 50 different small causes of TFR decline (X different policies, Y different technologies, Z different changes in culture etc.), and we will never be able to estimate them using the noisy data we have.
Agreed. You wrote:
> No article I’ve seen did a systematic comparison of different proposed variables, so it is somewhat difficult to judge the relative evidence for them.
The authors (perhaps in this paper, perhaps in discussion elsewhere) did a substantial review, precisely so as to isolate some variable that _could_ be judged accurately at all, in which the null hypothesis of mere noise did not dominate. Thus, you may be interested in their process/results.
I've seen so many conflicting hypotheses so it's nice to see such a sober analysis, thank you.
Do you have any thoughts on high fertility groups like the Amish and Haredem?
Religiousness is a strong correlate at the individual level, but country level data for religiousness is sparse/spotty. By request from another person, I tried out the dataset I could find with the most coverage. It does show small positive effects of religiousness, but given the small amount of data, it's not particularly convincing and the effect size is small. https://x.com/KirkegaardEmil/status/2043375308204568624
Seems to me it's a version of Baumol's cost effect. As the world gets richer in absolute terms, both the absolute cost (in education cost, child care, etc.) opportunity cost (of mom's lost earnings, career setback, etc.) of having a kid is higher than ever (as there are no productivity gains in child birthing or child rearing).
Another way to think about it - income vs wealth effect. We are wealthier than ever (which would predict higher consumption of child-having), but having a kid reduces your income so much relative to those who don't have kids (again both in absolute and relative terms) so much that it swamps the wealth effect.
You could also think of it the same way as with the obesity. People are eating too much, in part because there is just so much more diverse, tasty and cheap food around, delivered to your door using only your phone. It has never been easier to eat in other words, so people eat more. Similarly, life offers much entertainment and many things to try. It has never been easier to try new things, travel practically anywhere, live there, 1000s of obscure hobbies are showcased on the internet, and endless free entertainment. So relative to these life offerings, spending 9 months being pregnant and dealing with infants/toddlers for years doesn't seem so relatively appealing as it once did.
It's marriage rates which in turn are driven by male/female wage and status gaps: https://www.aporiamagazine.com/p/the-baby-boom. Contraception doesn't explain it (see Japan).
What variable would proxy feminism? Social cohesion? Amount of social liquidity?
There are several, none of which showed much of anything. Most obviously women in parliament %.
Why is Israeli TFR so high even among more secular jews? My guess is ultra-nationalism, ethnic paranoia, high overall religiosity and pro natalist policies, things any state could in theory emulate but are not ideologically kosher in any Western state.
Not for any reasons that could be discovered by this study.
Ukraine is in a far more exists risk situation then Israel and it hasn’t helped TFR. South Koreas response to the Korean War was to try and suppress its fertility.
Maybe part of the answer is that Israel provides a lot of financial support to families, has a good school system, and free IVF. They also have a low divorce rate for a modern western country (they don’t have no fault).
War and terror plays a role. I would guess, 5 years after Oct 7, 2023, fertility will be higher than just before Oct 7, 2023. Because there is fear that children (or later, adults) may die in a war.
For sure there's something to it. In some european countries the Baby Boom started when they were belligerents during WW2.
On the other hand ukrainian fertility is abysmal and worse than before the war. And I doubt the butchery in Gaza boosted palestinian fertility.
There is no butchery in Gaza. It's all islamist propaganda. Ukraine is interesting though. Maybe because it is for a lot of Ukrainians a war of choice that they haven't chosen.
I have been looking on the first plot (log vs normal) and I would have agreed with everything you said about it if the axis were switched!
For OECD countries it does not make difference either way. But the current plot suggests that for world for certain models the log variant has higher error (RMSE) than the non-log variant.
It even said so in the legend (below the line the log models are better).
Is the issue that you flipped the axis in the plot or you switched log and non-log in text? I guess the latter based on the intuitive reasoning, but I am not really sure.
Or am I misunderstanding something completely?
Are y out mixing Asians and western countries? I feel like these are almost separate universes.