Low birth rates in the past cause ageing and shrinking populations in the present, which cause low birth rates in the future. So negative population growth feeds on itself.
There are a variety of mechanisms by which this happens. (There may also be countervailing natural forces, as civilization is an extremely complex system with many means to regulate itself.) The main ones are:
1) An inverted population pyramid slows the pace at which the young gain status. Unless adequately adjusted for by society or culture, this raises the age of household formation, increasing the chance of medical issues or life disruption causing fertility shortfall. It also drives popular discontent and political polarization.
2) Shrinking populations reduce future demand and therefore current real investment opportunities. If the population is going to be smaller in future, the country doesn't need to build as many houses or apartments - or shops, or factories, or whatever. Meanwhile the cigar-shaped population pyramid attained during the demographic transition has peak supply of savings. This imbalance between desired investment and supply of real investment opportunities drives up asset prices, increasing the price of household formation.
3) Further, the investment / savings imbalance fosters the growth of, and makes it rational to invest in, a variety of bubbles and Ponzi schemes. This draws capital away from potentially productive uses (that could perhaps ameliorate the causes of the birth rate shortfall!)
4) In a democratic system, population pyramid inversion means the modal voter switches from being young to old and quite possibly from a net short position in financial assets (that is, desiring to buy a house in future) to long (that is, desiring to downsize or sell their house in future.) This leads to political pressure to support asset prices, which reinforces the effects of the investment / savings imbalance, and combined with the rising age of leadership ultimately makes the interests of the wealthy elderly politically all-powerful. In the extreme, in many cases, it becomes impossible to take actions (for example, building houses) which could ease the constraints that are lowering birth rates.
5) Cultural norms shift to normalize low birth rates. Families in media are depicted with fewer children and lower birth rates propagate by mimesis through real-life social networks. Required investment in children rises, infrastructure subtly becomes less convenient for children and those with children.
This is not intended to be an exhaustive explanation for falling birth rates. Certainly birth rates did not begin declining because of any of the mechanisms described here. This is an explanation of how birth rates have positive feedback and how, having declined, low birth rates cement themselves.
1) Status slowdown
In prehistory, the age distribution of the population would have been a neat pyramid. Lots of babies and infants: many of these, perhaps a third, would have died of sickness or malnutrition before reaching 5 years. Fewer children: perhaps a quarter of the five-year-olds would have died by 15. And fewer adults: another quarter of the 15-year-olds would die before 35 - of the men, many by violence, and of the women, many in childbirth. Maybe a third of total babies born would ultimately survive into their forties. Without medical technology, after 50 or so years hazard rates start to creep up again. There were very few elderly in a prehistoric society.
In such a world simply being old was a sign of fitness and value. While evolution is powered by randomness and accepts plenty of it, in a world without medicine the babies who survive to childhood will, on average, have better immune systems and fewer deleterious mutations than those who died in infancy. And those who survive to late adulthood will, on average, have demonstrated more competence at social coalition-building (and in men, at combat) than those who did not. And in humans, as physically weak animals who use knowledge and co-ordination to hunt larger stronger prey, rising age also means accumulation of knowledge and experience.
So the population pyramid would also, more or less, have been the status or command pyramid. A few senior veterans tell the larger group of prime-age workers what to do, who in turn boss around the more numerous junior workers, who train the children, who keep order over the scurrying throng of infants. From the youngest age children continually observe that those older than them are more knowledgeable and powerful (even literally, physically until the late teens) than them and therefore outrank them. Accordingly, it seems plausible that deference to seniority is innate in humans.
It is natural if not inevitable for social rank in humans - deference, respect, ability to tell others what to do - to be correlated with age. Even in chimpanzees and bonobos, older individuals are usually more senior in the local social hierarchy. This is not because age itself commands respect - it is because these are intelligent, social animals and both knowledge (of hunting or gathering spots, water sources, and so on) and social connections accumulate with time. In humans, as the most social and most co-operative of primates, the value of these assets is even more pronounced. Knowledge, connections, and wealth all compound over time. So typically older people will have more of each of these. And even when taken away from these sources of value (say, when travelling) humans typically gain comfort with themselves and therefore justified self-confidence - which is itself what commands respect - as they age.
In those times, the population would have been a constant pyramid, held down by extremely high rates of mortality. Women who survived to the end of their childbearing years might have had six or more children but - less than half of those survived to childbearing age themselves. Population growth in aggregate was incredibly slow, globally on the order of 0.01% annually.
Most humans who lived in prehistory had the experience of steadily rising in social rank and status until their typically untimely deaths. Everyone - if they demonstrated their fitness and their value by the mere fact of their survival so far - got to outrank a larger group of younger subordinates.
For most of the time since the Industrial Revolution, when humans broke free of Malthusian constraints, the population has been a neat pyramid. Deaths from sickness, childbirth, and violence all fell dramatically. But breaking free of Malthusian constraints meant that more and more people survived to adulthood to have children in turn, and therefore instead of an age distribution kept pyramid-shaped by high mortality the age distribution was kept pyramid shaped by increasingly large successive generations. By the early postwar era, women who survived to the end of their childbearing years might have had four or more children, and each of those would be on track to have four or more children themselves. So the human population exploded.
Today, for the first time in history, many societies no longer have a pyramid shaped age distribution with few seniors and more juniors, as since the 1960s birth rates in these societies have declined to substantially below replacement. So for the first time in history the population pyramid does not naturally map to a status pyramid. And it now appears that by the second half of this century the human population will be trending downwards for the first time in tens of thousands of years.
Let's go through an example. In 1970, the age distribution of the Japanese population from ages 21 to 85 formed a perfect pyramid. There were 2.4 million 21-year-olds; 1.7 million 31-year-olds; 1.5 million 41-year-olds; 1 million 51-year-olds; and so on. Imagine an archetypal hypothetical Japanese company - call it "Japan Inc" - whose population structure exactly mirrors that of Japan's, and which (for the purposes of simplicity) has four ranks in its company precisely by quartiles of seniority. The quarter of employees who are youngest and have least experience are the lowest rank, the second quartile by age and experience have the next rank, and so on. With the population pyramid of Japan in 1970, the youngest quartile by age / experience ends at 29, and so at 29 a worker is promoted from the lowest rank to the next. Then a worker reaches the median age of the company and is promoted again at 39. And a worker becomes top-quartile by seniority and reaches the highest of our four ranks at 52.
In 2023, the age distribution of the Japanese population from ages 21 to 85 forms an irregular spinning-top shape. There were 1.2 million 21-year-olds; 1.2 million 31-year-olds; 1.5 million 41-year-olds; 2 million 51-year-olds; 1.5 million 61-year-olds. Imagine our quartile-seniority-based Japan Inc again. Now, a worker is promoted from the lowest rank (after gaining more experience than a quarter of all workers there) not at 29 but at 39. A worker reaches median age and experience and is promoted again not at 39 but at 53. And finally reaches the top quartile not at 52 but at 69. In this toy, the number of years needed to be accumulated to rank up at each successive level have gone from 8 years, 11 years, 13 years in 1970 to 18 years, 14 years, and 17 years today. The distance between ladder rungs has increased.
Of course, rank systems in companies are far more complicated than that and (outside Japan) not purely seniority-based. But this cannot ameliorate the problem. As long as hierarchy is strongly correlated with age, and as long as we are social primates it will continue to be, then no matter how you arrange the rungs of the ladder it will take longer to move up the ladder in worlds with fewer junior workers than in worlds with more junior workers. And the age at which people attain top-ranked positions empirically appears to have risen significantly in the last few decades. The average age of US CEOs (https://corpgov.law.harvard.edu/2026/05/01/what-explains-the-rise-in-ceo-age/) rose from in 51 in 2000 to 61 in 2023. The average age of the US Senate rose from 53 in 1981 to 64 in 2026. Trump was the oldest first-time president in history when elected in 2016, and then Biden in 2020 and Trump in 2024 were in succession oldest president at inauguration in history. In Japan itself, the average age of company presidents has (surprisingly) risen only from 54 to 61 from 1990 to 2023.
Status is incredibly important for humans. Above all, it is important because it largely gates access to sex (sex at all for men, sex with the highest-ranking suitors for women); and sex, along with death, is one of the two most important things in the universe. What drives evolution drives everything.
Economists commonly claim that your archetypal ancient warlord would have given it all up for the fruits of modernity - plumbing, air-conditioning, antibiotics and painkillers, TikTok and video games. But the ancient warlord got to demand other people's wives and daughters (among many other, perhaps less satisfying, forms of status over others.) Some people would surely consider this worth more than technology. The logic of evolution suggests they are probably right.
The commodities and freedoms typically cited in economic analyses - food, which is freedom from hunger; shelter, which is freedom from cold and exposure; medicine, which is freedom from sickness and pain - are all freedoms from fear of death. Pain, hunger, and so on feel unpleasant to us because they are triggering our body's signals that we could be approaching death or permanent impairment of our bodily capital. This is why they are so self-evidently valuable and therefore so useful to talk about in toy economic stories. Sex, and therefore status, is not typically considered a proper subject for such analysis. But it is just as important.
The good reason why status cannot be analyzed economically is that it is zero-sum. More of it cannot be created in the short run. For someone to be higher status, someone else must be lower. Meanwhile, the probability of dying can always be lowered further, and feelings of pain, hunger, discomfort can always be reduced. Which is to say, (true) wealth can always continue to grow, until we are all immortal and feel only eternal bliss, which is likely an economic state that can approached asymptotically at best.
There's an issue here, which is that human utility - while created by evolution - somewhat diverges from true evolutionary utility. But in many contexts, small amounts of pain, hunger, and so on are in fact healthy, which is to say they increase the expected value of your ability to propagate your genes. With sex the issue is more pronounced. Sex feels pleasant to us because it triggers our body's signals that we could be approaching reproduction. But with our technology we have severed the link between sex and reproduction. Our brains can now get that utility without our genes getting utility. In many fields, not just reproduction, our technology has now advanced sufficiently to game the heuristics and targets evolution has set for us, which may become a problem for us.
But status can be created in the long run. The only way to add net supply of status is to add new low-status humans, with a promise to them that they will be high-status later, which is in turn achieved by adding new low-status humans. In other words: continuing population growth means growth in net supply of status. And continuing population growth is the only way to achieve growth in net supply of status.
This could be done with (low-status) immigration just as much as with new young workers. But negative population growth feeds on itself. In every country facing low birth rates, a large segment of the population feeling that the rungs on hierarchical ladders are now farther apart blame immigrants for taking their opportunities from them. This message plays well on - appears to be naturally amplified by the algorithms of - our collective information system, and has significant political force. So, as a stylized and very general rule, countries which face the worst problems with shrinking and ageing populations have the fewest immigrants. Tribalism is a deeply-rooted evolutionary impulse and probably it outright makes evolutionary sense: in an environment of limited resources, better to keep them for your closer genetic cousins than share them with your further ones. However, in today's world, where outright resource constraint is - while for some key resources tighter than it was 50 or 100 years ago - far more relaxed than it was in our evolutionary environment, and where for the first time (due to birth control) ever it is possible to have a shortage of young children, this instinct may be maladaptive.
2) Savings / investment imbalance
The fundamental driver of popular dissatisfaction in falling population societies is decreased availability of growth in status, as the distances between rungs on various hierarchical ladders grows on average. But an inverted population pyramid has more fundamental corrosive effects on the socioeconomy than just dissatisfaction with slow progress. It warps the balance of desired savings and investment opportunities in a manner that - from the perspective of a growing economy - impairs economic function and drives down population growth further in a positive feedback loop.
The first thing to understand is that almost all capital is complementary to humans. Not just to labor (which may be more or less entirely replaceable by capital soon) but humans as consumers. Obviously the classic economics examples of capital goods are complementary to labor by construction: a widget factory purchases one widgeting machine; using a widgeting machine, one worker can make one widget in one hour, and so on. But many more forms of capital than this archetype require humans to make them valuable. Today, in the US, privately owned real estate (land and structures) makes up a little under half of all private wealth. Consider a house: a home is only valuable if someone is willing to pay rent to live in it (or for the option value of being able to stay in it, in the case of vacation or second homes.) In unpopular rural areas of shrinking countries such as Japan and Italy, there are homes whose value has fallen to zero. Because no-one wants to live or stay in these areas. A home is granted value by human demand - not by the cost of construction materials or the labor that went into building it. In a future with very few humans and very many empty homes, these homes will be worthless.
People in a modern economy, more than any past economy, want or need to save for the future. It is only very recently in evolutionary history that humans frequently started living longer than they remained economically valuable. In advanced economies today the vast majority of the population actively expects to live many years past the end of their economically useful life and they need savings - enforceable claims on economically valuable assets, which in their dotage they can live off the returns from - to do this. Homes have historically been one of the main routes for such saving. Either directly, or at a smaller scale by saving in bank deposits which are lent out as mortgages, or similar intermediating transactions. With a growing population, all else equal, more houses will be needed in future than currently exist in the present. And so the desire of would-be retirees in the present to save by contributing to the building of new housing is matched by the desire of the larger future population to have more houses to live in. This applies not just to houses but to all forms of capital - to any variety of widgeting machine. If the population of a polity today is 10 million, and in 10 years it will be 11 million, enough homes to house 1 million people will need to be added to the housing stock in the meantime. Today enough shoe-making equipment to keep 10 million people's feet clad is sufficient, but in 20 years the polity will need enough shoe-making equipment to keep 11 million people in shoes. All else equal (in practice capital gets more efficient in many areas, and new forms of capital are continually invented) if the population is going to grow 10% enough investment will be required to increase the stock of all forms of capital by 10%.
Contrarily. If the future population is half the size of today's, all else equal the future economy needs half as many houses: half as many offices, shopping malls, shoemaking factories, guitar repair shops, you name it. Capital is complementary to humans not just because it needs complementary labor to produce output but primarily because it needs humans to provide demand to give its output value. (Of course, all else is never equal. When the widget factory is faced with a labor shortage, it upgrades to the new model of widget machine that can produce two widgets per worker-hour. Substitution and complementarity are inevitable. And so the square footage of residential space demanded per person to have a "respectable" home has steadily risen for decades. Since COVID-19, in city centers, it very much seems that the square footage demanded per person of office space to have a respectable professional office has increased enormously. It's easy to imagine that similar shifts will occur for other forms of commercial real estate as populations shrink.) But even the most fashionable form of capital today - GPUs and the intelligences that run on them - currently depend on humans to provide them with final demand, in subscriptions or per-token spending or in the form of consumer dollars that can be redirected by advertising.
It makes sense that if the future sees a shortage of labor relative to capital it would make sense to invest in capital that can substitute for, rather than be complementary to, humans. And, too, in the fullness of time AI may be economically emancipated and at last able to provide final demand of its own. My belief is that while this will ensure that "number go up" and will ameliorate the present imbalance between desired savings and opportunities for investment, it is very unlikely to reverse the feedback loop towards falling population and far more likely to represent the beginning of the final obsolescence of sapiens.
Let's go through an example. The US population was growing at over 1% a year in 1980. Assume this simply corresponds to 1% increase in required capital stock. The produced capital stock (structures, durable goods, etc.) of the US was valued at $10.5 trillion in 1980, about 3.5 times 1980 GDP of $2.9 trillion. So population growth alone implied an additional $100 billion in new investment - over 3% of GDP, nearly half of 1980's actual net savings of 7.1% of GDP. As capital stock deepening has apparently come to an end in advanced economies - which now appear to no longer need new capital stock, merely to regularly replace old capital stock as it obsolesces and depreciates - capital stock growth needed to support US population growth of about 0.5% now accounts for more than all of the US's 1.2% net savings rate (the current capital stock to GDP ratio is about 3.4 times.)
Today, Japan's annual population growth rate is around -0.5% a year. In 1970 it was 1.3%. This swing of 1.8% is about as much as can be expected over the course of a completed birth rate decline. Assuming for a rough estimate that the produced capital stock to GDP ratio is usually about 3.5 times - in Japan, currently, it is estimated at 3.7x - it corresponds to a fall in required investment of over 6% of GDP. Assuming desired savings behavior remains unchanged (which of course, it will not be) this could plausibly drive quite a big price increase in long-dated investment assets and presumably fall in yields.
However, while savings behavior is of course hard to understand - driven as it is by culture, incentives, technology, available or apparent returns, and more - it seems much more likely that desired savings (all else, such as price, equal) increases during the course of a birth rate decline. Because birth rate decline inverts the population pyramid which, after a few decades, causes the economy to contain a higher proportion of more senior, higher-paid, possibly higher-saving workers compared to more junior, lower-paid, possibly lower-saving workers. The economic changes brought on by birth rate decline will change the typical pay structure by age, and the desired savings by age, too, in unpredictable ways. But for an indication, consider statically: in the US, according to the 2024 BLS Consumer Expenditure Survey, the average income for a "consumer unit" (i.e. household) headed by someone aged 25-34 was $102,000. For the peak average income age bracket, 45-54, it was $141,000. And income exceeded expenditures for 25-34 year olds by 27%, while for 45-54 year olds it was 29% - note that standard lifecycle consumption hypothesis models would typically assume a larger increase in desired savings with age than that. Applying the 2024 Consumer Expenditure Survey ratios of income and expenditures by age bracket to the population structure of the US in 1980 and 2025 would imply an increase in the desired savings rate since 1980 of 0.8%.
3) Bubbles
Increased savings (and plausibly, decreased elasticity of savings to yields) and decreased need for investment cause the prices of investment assets to rise. They seem to create new zero-or-negative-yielding investment assets, by creating digital assets and by turning collectibles, art, fine wine, and so on into "asset classes". They create a need in the market for hype companies which promise some futuristic technology but cannot actually deliver it.
Arguably, a large enough demographic shift in the supply of and demand for investment opportunities changes the nature of the economy, and therefore the function of the banking system and central bank, from investment-driven to savings-driven. When we think of economic history, especially US economic history, we think of an economy with rapidly expanding credit and real assets, with many possible new investment opportunities clamoring for savings to be invested in them, leading to high yields on assets and many productive opportunities going unfunded. In turn this led to a credit system driven by a vigorous search for the best investment opportunities so as to maximize scarce savings, and a central bank whose purpose was to serve as lender of last resort if or when this vigorous search got overleveraged and had an accident. Such a central bank would not need to pay interest on its liabilities and would charge a high rate to lend, effectively setting a floor (zero) and ceiling (its intervention lending rate) on all good credit rates in the economy. But a savings-driven economy has stagnant or shrinking real assets, with few possible new investment opportunities, with low yields on assets and many unproductive opportunities getting funded. It has a credit system driven by a desperate search for any plausible investment opportunities so as to deploy excess capital, and a central bank whose purpose is to serve as borrower of last resort, manufacturing sufficiently high-yielding liabilities in which other actors can place excess funds, setting a floor (its borrowing rate) on all credit rates in the economy.
This excess of savings and shortage of new investment opportunities should massively increase the price of investment assets. Not only that, but it creates bubbles. As the unproductive zero-yielding assets created by this shift to a savings-driven economy rise in price, especially if they outperform actual yielding assets, they gain speculative momentum and may become rational to invest in in the short-run. And so it is that an enormous variety of non-yielding collectibles have risen enormously in price in recent decades. A Black Lotus from the nerd card game Magic The Gathering (even the 1993 Unlimited edition, the cheapest by far of the three official tournament-legal Black Lotus printings) sold for $1,057 in December 2010 and $17,050 in May 2026. The Rare Whisky Icon 100 price index of valuable whiskies rose from 319 at inception in 2016 to a peak of 923 in 2022 (and is since down to 550 in June 2026); the WatchCharts Rolex Market Index (for secondary market Rolex prices) similarly went from 16,000 at inception in 2017 to a peak of 41,000 in 2022 and is since down to 29,000 in June 2026. The 1995 Lego spaceship set Star Hawk II - which I received from my grandmother for my 9th birthday, one of my most beloved childhood gifts - in June 2026 now sells, if sealed and complete, for an estimated $488, up from an original RRP of $26.99. Glove Cases for the video game Counter Strike 2 sold for $0.01 throughout 2018, and on June 7 2026 the best bid is $23.09; the Chroma Case for the same game traded from $0.04 to $0.09 throughout 2018 and now have a best bid of $6.99; Operation Hydra Cases reached a peak price of $3.96 in 2018 and right now sell for $50.90. There are endless more examples, from handbags through classic cars and so on, but the best examples are cryptocurrencies. They never achieved widespread use as an alternate means of exchange - except for in crime, which is indeed a big market - but as their prices rose a significant industry built up around them to convince people to invest in them, to develop new cryptocurrencies, and so on, which allowed them to achieve some use purely as a (mostly fictitious) store of wealth.
These also illustrate most clearly how these bubbles can draw talent and initiative away from other potentially productive activities. Many of the most skilled software developers of the 2010s went to work in crypto, and it was rational for them to do so, and they were rewarded highly for it. And, today, even among crypto fans the crypto world is known to have far more than its fair share of grifters, scammers, and deluded zealots, because bubbles driven by an excess of demand to save over real investment opportunities create a profitable market niche for these characters. If extended, such bubbles encourage more people to adopt more aspects of grifter and scammer into their own personas and ultimately, if extended for long enough, presumably increase the frequency of genetic alleles linked with such behavior. While not as prominent and widespread as in crypto, there are plenty of examples of such characters in public market equities (and perhaps in politics) too. Negative population growth driving bubbles and therefore grift is just one of many ways in which negative growth incentivizes negative-sum behavior on the part of actors in the system.
Many researchers have considered the likely effects of demographic change on economic variables and asset prices - Arnott and Chavez, Juselius and Takats, Goodhart and Pradhan - but to crudely summarize their work, the simplest (and therefore likely most accurate) model of demographic effects is that higher dependent shares cause higher inflation and lower asset prices, while lower dependent shares cause lower inflation and higher asset prices. Because low dependent shares mean a more productive economy (as more of the population is able to participate in the economy) and more savings potential, while high dependent shares require dissaving to support them and mean less of the population is able to contribute.
And some part of the explanation for low inflation and high asset valuations, on average, in recent decades is that dependent shares fell to historically low levels and are now only rising slowly. In fact, in most countries with low birth rates, they will rise very slowly and even with a rising elderly proportion of the population will not rise to historically high levels.
The reason is: not just the elderly, but also children, are (expensive, economically burdensome) dependents. While a birth rate decline eventually inverts population pyramids and leads to an excess of elderly dependents, it immediately reduces the burden of young dependents. The aged economies of the future, in most reasonable projections, will not actually have a much higher dependent share than the young economies of the past. During China's demographic transition, from 1980 to the early 2020s, its total young and old (<20 and >65) share of the population fell from 50% to 35% while its birth rate fell from 2.7 children per woman to 1. Now that it has 1 child per woman - its total young and old share of the population is forecast to rise from 35% back to 50% by 2080. The US' total young and old dependent share is forecast to rise from 41% today to 46% in 2080; Japan's from 45% today to 50% in 2080. For comparison, the US' total young and old dependent share was 48% in 1960.
What this means is that we will be no (or only barely) more burdened by elderly care and elderly nappy-changing than we were in the past by childcare and baby nappy-changing - as long as we remain within the comforting spiral of demographic decline. It is only when or if birth rates rise again that total dependency ratios will get much higher than were seen in the past.
This may be the most important mechanism of all. Falling birth rates are, in the short run, economically very pleasant. They immediately reduce the dependency burden, freeing up labor and capital for other uses, driving up short-run growth. If they have negative effects these do not kick in for decades. It is easy to imagine both governments and market systems being subtly incentivized to bring them about. But rising birth rates are, in the short run, economically painful. Once polities have gotten used to low birth rates it is hard to imagine - as much as politicians might overtly claim to wish to increase birth rates - that the negative economic effects of a major increase in birth rates could be tolerated.
4) Political economy
The inversion of the population pyramid also drives a change in the political economy of asset prices. A population with a majority of young voters is indifferent to asset prices or prefers lower asset prices. A population with a majority of old voters prefers high asset prices. As voters get their way and asset prices rise, the resulting rise in wealth inequality exacerbates the extent to which effective political power speaks for asset owners. This ultimately leads not just to high housing (and bond, and equity, and gold) prices but to pro-incumbent policy: with with less enforcement of antitrust, monopoly and so on, regulations that stifle potential new competitors to existing incumbents, a weakening of the political power of labor, and ultimately much higher profit margins; with restrictions and delays on building new housing and real assets and infrastructure. In a democracy, an inverted population pyramid leads to more voting power for incumbent capital owners (who are likely to be old). It is in the interest of incumbent capital owners not to allow the creation of new capital anywhere the new capital could be competitive with theirs. This inability to create new real capital - houses and so on - prevents births, cementing low fertility and an inverted population pyramid.
According to US Census data (historically broken into age brackets 18-24, 25-44, 45-64, and 65+) the 1984 election had 53 million voters aged 44 or under, and 49 million voters aged 45 or older. In the 2024 election, there were 60 million voters aged 44 or under, and 95 million voters aged 45 or older. Youth participation has actually risen over that time (by 3.1% for the 18-24 year-old age group) but by less than elder participation (with the proportion of 65+ year olds voting rising by 4.0%.)
Older voters are more likely to own homes, and much more likely to own multiple homes. They are more likely to own equity, and much more likely to own lots of equity. In the extreme (65+ voters went from being 18% of the vote to 28% of the vote from 1984 to 2024), they are much more likely to derive all of their income or consumption from assets and not from labor. Of course, they also receive benefits which make up an increasingly large share of the economy and of federal tax receipts. Social Security plus Medicare went from 5.9% of US GDP in 1984 to 8.0% in 2024 - from 35% of US federal tax income to 47%. The government deficits this has driven have, in turn, been allowed by the global excess of savings resulting from ongoing demographic transition. The increasing share of the voter base receiving these transfers has become a powerful special interest group dedicated to these transfers. As these transfers become an increasing share of the government budget they reduce fiscal and political headroom for measures that could increase the birth rate.
The extent to which local governments (dominated by elderly participation) have blocked new housing development in key economic hubs has been widely remarked on. Across the US (according to Federal Reserve data series) residential property prices have risen 14.4 times since 1975, while commercial property prices have risen only 6 times. A large part, if not all, of that outperformance will be the rise in the value of the rent attached to the legal right for a property to be residential. Permitting and connecting energy generation, and new development of natural resources, have gotten harder too. It is possible that personality changes associated with ageing - rising risk aversion, or rising aversion to change - may directly motivate some of this as well as economic incentives. This limitation on actual physical growth - not measured economic growth, but growth in actual energy, materials, living space - obviously inhibits birth rates.
Meanwhile, increasing concentration of capital, and perhaps the increasing effectiveness of mass public manipulation technologies in the information age, have led to many political systems, especially the US, moving generally closer to one dollar one vote than one man one vote. (In the US this is exemplified by Citizens United vs FEC, but actual election spending has risen a surprisingly small amount in the past few decades.) Ultimately, it seems, while individual firms compete with each other for labor and for consumers, in aggregate the negotiation between capital and labor over the division between the fruits of their joint product is - a pure political struggle. And in monopolistic markets so too is the rent that capital can take from consumers. While in the past the median voter expected to have more future labor income than they then had capital, today the median voter has more capital than labor income. So the median voter now benefits when corporations take more than what had previously been their fair share from labor and from consumers. And so the government has tacitly encouraged profit margins to rise to historically unprecedented levels (after-tax profits were 9.2% of gross domestic income in 2024, above the pre-2020 record of 8.9% set in 1929, and far above the 1945-2000 average of 5.3%), justifying the price increase in US equities that was driven by the savings / investment imbalance, further entrenching the wealth and thus political power of US equity owners. To some extent this suppression of wages in favor of return on capital - which in aggregate transfers income from young to old - inhibits birth rates over time, but it may support birth rates in the short run while the bubble is rapidly expanding. South Korea (with the lowest birth rate of any major country in the world) saw its birth rate rise 16% from the first quarter of 2025 to the first quarter of 2026 - after a year in which its equity market rose 93% in dollar terms.
While all the mechanisms described above "make sense" - should clearly work, in isolation - their effects are refracted through the endless complexity of the economy. While the demographic transition is happening across most of the world - all but the very least developed countries, not just advanced economies but all economies with effective trade and cultural links to advanced economies - excess savings are more pronounced in some countries than others and different countries have different relationships with the imbalances caused by the demographic transition. In practice, in the 2020s, excess savings are above all concentrated in China, whose government forces its people to save in this manner and attempts to direct those savings into continuous massive capital expansion. A small portion of those savings leak out and overwhelm already tight markets in the rest of the world's supply of investment opportunities. Simultaneously China keeps its currency significantly undervalued and uses its rapidly expanding real capital base to dump on export markets globally.
Excess investment is above all concentrated in the US, which - as the world's reserve currency, largest consumer market, and cultural hegemon - is naturally positioned to host the new natural monopolies (search, social networking, spreadsheets) that arose with the information age. Whether LLM-type AI will turn out to be a natural monopoly too remains to be seen. As US assets have been steadily pushed up by this imbalance, they have attracted momentum flows and gone into a bubble. As a result the US has become generally uncompetitive, except for a few superstar industries able to use the high price of US assets to attract the world's most ambitious people.
A number of other countries, mostly China's neighbors, have followed China in having a low-rate, low currency, financially repressive policy. While it makes sense for countries facing most rapid demographic decline to encourage their residents to accumulate foreign assets, we can perhaps unfairly stereotype the situation as: the residents of these countries work extremely hard, never seeing the fruits of their labor, to provide us with most of our material goods and to transfer vast profits to US corporations, in exchange for tokens we conjured out of air, and any disruption to this state of affairs will lead to us devaluing their hard-earned tokens. On the other hand, it has allowed them to become more advanced in most forms of technology and manufacturing, while the US suffers from a terrible form of Dutch disease, sitting as it does on an infinite supply of the most valuable natural resource in the world, reserve currency status.
5) Culture
So in aggregate, the economic imbalances induced by below-replacement birth rates drive fertility further down - ultimately, in essence, because driving fertility further down is the easiest way to prevent already-high dependency ratios from worsening further. But low fertility reinforces itself culturally as well as economically. Above all, humans are mimetic and easily influenced by social norms. Marriages and childbirths appear to be contagious within peer groups. To be normal is, always, aspirational. So in societies with high birth rates people will naturally learn from childhood that high birth rates are normal, healthy, and good. They learn this from their families, friends, and communities more than from media - although today as humans are less strongly socially connected, and have increasingly strong parasocial relationships via media, media may be more effective than it was in the past. Today, while children are (generally) viewed as aspirational, larger families are increasingly uncommon and much later age of marriage has become normalized or favored in advanced economies, and this must lower birth rates somewhat in the short run (as it spreads births out into the future) and slightly in the long run (as the risk of failing to marry or of later medical, etc., problems preventing children increases).
Of course, fundamentally, as humans in a physical world sharing various kinds of real infrastructure, as fewer people have children the cost of having children rises subtly in various ways. There are fewer daycares around, people judge you more when the baby cries on a plane, you have to turn down more social invitations as they would be unwelcoming to children. That kind of thing. But the cost of having children has risen in various more direct and hard-to-understand ways which may be linked to the fertility feedback mechanisms described here. Education and childcare have become enormously more expensive, rising much faster than inflation in most advanced countries (this may be purely because of the Baumol effect). Harder to quantify but universally acknowledged, the time and effort cost of children has risen significantly, especially in the US. Upper-class children require far more time-(and money-)intensive activities to culture them sufficiently to compete in their admissions arms races; all children now require permanent carers, where once they could be left to fend for themselves. Parents who let children venture out or play on their own are now considered willfully negligent and are at risk of serious consequences. This adds multiple hours per day to the time cost of children - leading to the plausible argument that children have now become a superior leisure good. This stigmatization of unsupervised freedom for children quite likely makes them less well equipped for adult life than they would have been otherwise.
Rising safety concern for children drives laws that directly increase the price of children as well as increases their time and attention cost. The famous example is child car safety seat legislation, which economists determined drove a 0.7% reduction in birth rates for families with two existing children - who would thus likely have to upgrade to a bigger car if needing a child safety seat - compared to similar families in states without child safety seat laws. But examples will abound: window guards, daycare regulations, playground regulations... These rules surely do marginally make children safer. But many of them will have a larger marginal impact on birth rates.
In short, as birth rates have fallen, children have become more and more valuable and thus more and more expensive. And cultural standards have shifted towards a low-fertility norm. It is possible to imagine these trends being reversed by determined policy effort but, for now, easier to imagine them continuing.
Conclusion
Low birth rates cause low birth rates. They do this economically, politically, and culturally. They cause economic and social pressures that raise the price (in money and time) of children, reduce living space and real resources available for children, and ultimately lead towards an economic state in which the short-run dissaving or dependency pressure of additional kids becomes intolerable.
Enormously more could be said about many factors which contribute to fertility rates - some of which, through the complexity of the economy, may actually bear somewhat on the mechanisms described here and their interactions. But the argument over what has caused the fertility decline is interminable and there is no-one left to convince, all commentators have already picked their sides. This is my best brief effort to explain why low fertility matters: because low fertility begets low fertility, and thus becomes a trap that will take a remarkable shock or incredible policy effort to break out of.
Low birth rates in the past cause ageing and shrinking populations in the present, which cause low birth rates in the future. So negative population growth feeds on itself.
There are a variety of mechanisms by which this happens. (There may also be countervailing natural forces, as civilization is an extremely complex system with many means to regulate itself.) The main ones are:
1) An inverted population pyramid slows the pace at which the young gain status. Unless adequately adjusted for by society or culture, this raises the age of household formation, increasing the chance of medical issues or life disruption causing fertility shortfall. It also drives popular discontent and political polarization.
2) Shrinking populations reduce future demand and therefore current real investment opportunities. If the population is going to be smaller in future, the country doesn't need to build as many houses or apartments - or shops, or factories, or whatever. Meanwhile the cigar-shaped population pyramid attained during the demographic transition has peak supply of savings. This imbalance between desired investment and supply of real investment opportunities drives up asset prices, increasing the price of household formation.
3) Further, the investment / savings imbalance fosters the growth of, and makes it rational to invest in, a variety of bubbles and Ponzi schemes. This draws capital away from potentially productive uses (that could perhaps ameliorate the causes of the birth rate shortfall!)
4) In a democratic system, population pyramid inversion means the modal voter switches from being young to old and quite possibly from a net short position in financial assets (that is, desiring to buy a house in future) to long (that is, desiring to downsize or sell their house in future.) This leads to political pressure to support asset prices, which reinforces the effects of the investment / savings imbalance, and combined with the rising age of leadership ultimately makes the interests of the wealthy elderly politically all-powerful. In the extreme, in many cases, it becomes impossible to take actions (for example, building houses) which could ease the constraints that are lowering birth rates.
5) Cultural norms shift to normalize low birth rates. Families in media are depicted with fewer children and lower birth rates propagate by mimesis through real-life social networks. Required investment in children rises, infrastructure subtly becomes less convenient for children and those with children.
This is not intended to be an exhaustive explanation for falling birth rates. Certainly birth rates did not begin declining because of any of the mechanisms described here. This is an explanation of how birth rates have positive feedback and how, having declined, low birth rates cement themselves.
1) Status slowdown
In prehistory, the age distribution of the population would have been a neat pyramid. Lots of babies and infants: many of these, perhaps a third, would have died of sickness or malnutrition before reaching 5 years. Fewer children: perhaps a quarter of the five-year-olds would have died by 15. And fewer adults: another quarter of the 15-year-olds would die before 35 - of the men, many by violence, and of the women, many in childbirth. Maybe a third of total babies born would ultimately survive into their forties. Without medical technology, after 50 or so years hazard rates start to creep up again. There were very few elderly in a prehistoric society.
In such a world simply being old was a sign of fitness and value. While evolution is powered by randomness and accepts plenty of it, in a world without medicine the babies who survive to childhood will, on average, have better immune systems and fewer deleterious mutations than those who died in infancy. And those who survive to late adulthood will, on average, have demonstrated more competence at social coalition-building (and in men, at combat) than those who did not. And in humans, as physically weak animals who use knowledge and co-ordination to hunt larger stronger prey, rising age also means accumulation of knowledge and experience.
So the population pyramid would also, more or less, have been the status or command pyramid. A few senior veterans tell the larger group of prime-age workers what to do, who in turn boss around the more numerous junior workers, who train the children, who keep order over the scurrying throng of infants. From the youngest age children continually observe that those older than them are more knowledgeable and powerful (even literally, physically until the late teens) than them and therefore outrank them. Accordingly, it seems plausible that deference to seniority is innate in humans.
It is natural if not inevitable for social rank in humans - deference, respect, ability to tell others what to do - to be correlated with age. Even in chimpanzees and bonobos, older individuals are usually more senior in the local social hierarchy. This is not because age itself commands respect - it is because these are intelligent, social animals and both knowledge (of hunting or gathering spots, water sources, and so on) and social connections accumulate with time. In humans, as the most social and most co-operative of primates, the value of these assets is even more pronounced. Knowledge, connections, and wealth all compound over time. So typically older people will have more of each of these. And even when taken away from these sources of value (say, when travelling) humans typically gain comfort with themselves and therefore justified self-confidence - which is itself what commands respect - as they age.
In those times, the population would have been a constant pyramid, held down by extremely high rates of mortality. Women who survived to the end of their childbearing years might have had six or more children but - less than half of those survived to childbearing age themselves. Population growth in aggregate was incredibly slow, globally on the order of 0.01% annually.
Most humans who lived in prehistory had the experience of steadily rising in social rank and status until their typically untimely deaths. Everyone - if they demonstrated their fitness and their value by the mere fact of their survival so far - got to outrank a larger group of younger subordinates.
For most of the time since the Industrial Revolution, when humans broke free of Malthusian constraints, the population has been a neat pyramid. Deaths from sickness, childbirth, and violence all fell dramatically. But breaking free of Malthusian constraints meant that more and more people survived to adulthood to have children in turn, and therefore instead of an age distribution kept pyramid-shaped by high mortality the age distribution was kept pyramid shaped by increasingly large successive generations. By the early postwar era, women who survived to the end of their childbearing years might have had four or more children, and each of those would be on track to have four or more children themselves. So the human population exploded.
Today, for the first time in history, many societies no longer have a pyramid shaped age distribution with few seniors and more juniors, as since the 1960s birth rates in these societies have declined to substantially below replacement. So for the first time in history the population pyramid does not naturally map to a status pyramid. And it now appears that by the second half of this century the human population will be trending downwards for the first time in tens of thousands of years.
Let's go through an example. In 1970, the age distribution of the Japanese population from ages 21 to 85 formed a perfect pyramid. There were 2.4 million 21-year-olds; 1.7 million 31-year-olds; 1.5 million 41-year-olds; 1 million 51-year-olds; and so on. Imagine an archetypal hypothetical Japanese company - call it "Japan Inc" - whose population structure exactly mirrors that of Japan's, and which (for the purposes of simplicity) has four ranks in its company precisely by quartiles of seniority. The quarter of employees who are youngest and have least experience are the lowest rank, the second quartile by age and experience have the next rank, and so on. With the population pyramid of Japan in 1970, the youngest quartile by age / experience ends at 29, and so at 29 a worker is promoted from the lowest rank to the next. Then a worker reaches the median age of the company and is promoted again at 39. And a worker becomes top-quartile by seniority and reaches the highest of our four ranks at 52.
In 2023, the age distribution of the Japanese population from ages 21 to 85 forms an irregular spinning-top shape. There were 1.2 million 21-year-olds; 1.2 million 31-year-olds; 1.5 million 41-year-olds; 2 million 51-year-olds; 1.5 million 61-year-olds. Imagine our quartile-seniority-based Japan Inc again. Now, a worker is promoted from the lowest rank (after gaining more experience than a quarter of all workers there) not at 29 but at 39. A worker reaches median age and experience and is promoted again not at 39 but at 53. And finally reaches the top quartile not at 52 but at 69. In this toy, the number of years needed to be accumulated to rank up at each successive level have gone from 8 years, 11 years, 13 years in 1970 to 18 years, 14 years, and 17 years today. The distance between ladder rungs has increased.
Of course, rank systems in companies are far more complicated than that and (outside Japan) not purely seniority-based. But this cannot ameliorate the problem. As long as hierarchy is strongly correlated with age, and as long as we are social primates it will continue to be, then no matter how you arrange the rungs of the ladder it will take longer to move up the ladder in worlds with fewer junior workers than in worlds with more junior workers. And the age at which people attain top-ranked positions empirically appears to have risen significantly in the last few decades. The average age of US CEOs (https://corpgov.law.harvard.edu/2026/05/01/what-explains-the-rise-in-ceo-age/) rose from in 51 in 2000 to 61 in 2023. The average age of the US Senate rose from 53 in 1981 to 64 in 2026. Trump was the oldest first-time president in history when elected in 2016, and then Biden in 2020 and Trump in 2024 were in succession oldest president at inauguration in history. In Japan itself, the average age of company presidents has (surprisingly) risen only from 54 to 61 from 1990 to 2023.
Status is incredibly important for humans. Above all, it is important because it largely gates access to sex (sex at all for men, sex with the highest-ranking suitors for women); and sex, along with death, is one of the two most important things in the universe. What drives evolution drives everything.
Economists commonly claim that your archetypal ancient warlord would have given it all up for the fruits of modernity - plumbing, air-conditioning, antibiotics and painkillers, TikTok and video games. But the ancient warlord got to demand other people's wives and daughters (among many other, perhaps less satisfying, forms of status over others.) Some people would surely consider this worth more than technology. The logic of evolution suggests they are probably right.
The commodities and freedoms typically cited in economic analyses - food, which is freedom from hunger; shelter, which is freedom from cold and exposure; medicine, which is freedom from sickness and pain - are all freedoms from fear of death. Pain, hunger, and so on feel unpleasant to us because they are triggering our body's signals that we could be approaching death or permanent impairment of our bodily capital. This is why they are so self-evidently valuable and therefore so useful to talk about in toy economic stories. Sex, and therefore status, is not typically considered a proper subject for such analysis. But it is just as important.
The good reason why status cannot be analyzed economically is that it is zero-sum. More of it cannot be created in the short run. For someone to be higher status, someone else must be lower. Meanwhile, the probability of dying can always be lowered further, and feelings of pain, hunger, discomfort can always be reduced. Which is to say, (true) wealth can always continue to grow, until we are all immortal and feel only eternal bliss, which is likely an economic state that can approached asymptotically at best.
There's an issue here, which is that human utility - while created by evolution - somewhat diverges from true evolutionary utility. But in many contexts, small amounts of pain, hunger, and so on are in fact healthy, which is to say they increase the expected value of your ability to propagate your genes. With sex the issue is more pronounced. Sex feels pleasant to us because it triggers our body's signals that we could be approaching reproduction. But with our technology we have severed the link between sex and reproduction. Our brains can now get that utility without our genes getting utility. In many fields, not just reproduction, our technology has now advanced sufficiently to game the heuristics and targets evolution has set for us, which may become a problem for us.
But status can be created in the long run. The only way to add net supply of status is to add new low-status humans, with a promise to them that they will be high-status later, which is in turn achieved by adding new low-status humans. In other words: continuing population growth means growth in net supply of status. And continuing population growth is the only way to achieve growth in net supply of status.
This could be done with (low-status) immigration just as much as with new young workers. But negative population growth feeds on itself. In every country facing low birth rates, a large segment of the population feeling that the rungs on hierarchical ladders are now farther apart blame immigrants for taking their opportunities from them. This message plays well on - appears to be naturally amplified by the algorithms of - our collective information system, and has significant political force. So, as a stylized and very general rule, countries which face the worst problems with shrinking and ageing populations have the fewest immigrants. Tribalism is a deeply-rooted evolutionary impulse and probably it outright makes evolutionary sense: in an environment of limited resources, better to keep them for your closer genetic cousins than share them with your further ones. However, in today's world, where outright resource constraint is - while for some key resources tighter than it was 50 or 100 years ago - far more relaxed than it was in our evolutionary environment, and where for the first time (due to birth control) ever it is possible to have a shortage of young children, this instinct may be maladaptive.
2) Savings / investment imbalance
The fundamental driver of popular dissatisfaction in falling population societies is decreased availability of growth in status, as the distances between rungs on various hierarchical ladders grows on average. But an inverted population pyramid has more fundamental corrosive effects on the socioeconomy than just dissatisfaction with slow progress. It warps the balance of desired savings and investment opportunities in a manner that - from the perspective of a growing economy - impairs economic function and drives down population growth further in a positive feedback loop.
The first thing to understand is that almost all capital is complementary to humans. Not just to labor (which may be more or less entirely replaceable by capital soon) but humans as consumers. Obviously the classic economics examples of capital goods are complementary to labor by construction: a widget factory purchases one widgeting machine; using a widgeting machine, one worker can make one widget in one hour, and so on. But many more forms of capital than this archetype require humans to make them valuable. Today, in the US, privately owned real estate (land and structures) makes up a little under half of all private wealth. Consider a house: a home is only valuable if someone is willing to pay rent to live in it (or for the option value of being able to stay in it, in the case of vacation or second homes.) In unpopular rural areas of shrinking countries such as Japan and Italy, there are homes whose value has fallen to zero. Because no-one wants to live or stay in these areas. A home is granted value by human demand - not by the cost of construction materials or the labor that went into building it. In a future with very few humans and very many empty homes, these homes will be worthless.
People in a modern economy, more than any past economy, want or need to save for the future. It is only very recently in evolutionary history that humans frequently started living longer than they remained economically valuable. In advanced economies today the vast majority of the population actively expects to live many years past the end of their economically useful life and they need savings - enforceable claims on economically valuable assets, which in their dotage they can live off the returns from - to do this. Homes have historically been one of the main routes for such saving. Either directly, or at a smaller scale by saving in bank deposits which are lent out as mortgages, or similar intermediating transactions. With a growing population, all else equal, more houses will be needed in future than currently exist in the present. And so the desire of would-be retirees in the present to save by contributing to the building of new housing is matched by the desire of the larger future population to have more houses to live in. This applies not just to houses but to all forms of capital - to any variety of widgeting machine. If the population of a polity today is 10 million, and in 10 years it will be 11 million, enough homes to house 1 million people will need to be added to the housing stock in the meantime. Today enough shoe-making equipment to keep 10 million people's feet clad is sufficient, but in 20 years the polity will need enough shoe-making equipment to keep 11 million people in shoes. All else equal (in practice capital gets more efficient in many areas, and new forms of capital are continually invented) if the population is going to grow 10% enough investment will be required to increase the stock of all forms of capital by 10%.
Contrarily. If the future population is half the size of today's, all else equal the future economy needs half as many houses: half as many offices, shopping malls, shoemaking factories, guitar repair shops, you name it. Capital is complementary to humans not just because it needs complementary labor to produce output but primarily because it needs humans to provide demand to give its output value. (Of course, all else is never equal. When the widget factory is faced with a labor shortage, it upgrades to the new model of widget machine that can produce two widgets per worker-hour. Substitution and complementarity are inevitable. And so the square footage of residential space demanded per person to have a "respectable" home has steadily risen for decades. Since COVID-19, in city centers, it very much seems that the square footage demanded per person of office space to have a respectable professional office has increased enormously. It's easy to imagine that similar shifts will occur for other forms of commercial real estate as populations shrink.) But even the most fashionable form of capital today - GPUs and the intelligences that run on them - currently depend on humans to provide them with final demand, in subscriptions or per-token spending or in the form of consumer dollars that can be redirected by advertising.
It makes sense that if the future sees a shortage of labor relative to capital it would make sense to invest in capital that can substitute for, rather than be complementary to, humans. And, too, in the fullness of time AI may be economically emancipated and at last able to provide final demand of its own. My belief is that while this will ensure that "number go up" and will ameliorate the present imbalance between desired savings and opportunities for investment, it is very unlikely to reverse the feedback loop towards falling population and far more likely to represent the beginning of the final obsolescence of sapiens.
Let's go through an example. The US population was growing at over 1% a year in 1980. Assume this simply corresponds to 1% increase in required capital stock. The produced capital stock (structures, durable goods, etc.) of the US was valued at $10.5 trillion in 1980, about 3.5 times 1980 GDP of $2.9 trillion. So population growth alone implied an additional $100 billion in new investment - over 3% of GDP, nearly half of 1980's actual net savings of 7.1% of GDP. As capital stock deepening has apparently come to an end in advanced economies - which now appear to no longer need new capital stock, merely to regularly replace old capital stock as it obsolesces and depreciates - capital stock growth needed to support US population growth of about 0.5% now accounts for more than all of the US's 1.2% net savings rate (the current capital stock to GDP ratio is about 3.4 times.)
Today, Japan's annual population growth rate is around -0.5% a year. In 1970 it was 1.3%. This swing of 1.8% is about as much as can be expected over the course of a completed birth rate decline. Assuming for a rough estimate that the produced capital stock to GDP ratio is usually about 3.5 times - in Japan, currently, it is estimated at 3.7x - it corresponds to a fall in required investment of over 6% of GDP. Assuming desired savings behavior remains unchanged (which of course, it will not be) this could plausibly drive quite a big price increase in long-dated investment assets and presumably fall in yields.
However, while savings behavior is of course hard to understand - driven as it is by culture, incentives, technology, available or apparent returns, and more - it seems much more likely that desired savings (all else, such as price, equal) increases during the course of a birth rate decline. Because birth rate decline inverts the population pyramid which, after a few decades, causes the economy to contain a higher proportion of more senior, higher-paid, possibly higher-saving workers compared to more junior, lower-paid, possibly lower-saving workers. The economic changes brought on by birth rate decline will change the typical pay structure by age, and the desired savings by age, too, in unpredictable ways. But for an indication, consider statically: in the US, according to the 2024 BLS Consumer Expenditure Survey, the average income for a "consumer unit" (i.e. household) headed by someone aged 25-34 was $102,000. For the peak average income age bracket, 45-54, it was $141,000. And income exceeded expenditures for 25-34 year olds by 27%, while for 45-54 year olds it was 29% - note that standard lifecycle consumption hypothesis models would typically assume a larger increase in desired savings with age than that. Applying the 2024 Consumer Expenditure Survey ratios of income and expenditures by age bracket to the population structure of the US in 1980 and 2025 would imply an increase in the desired savings rate since 1980 of 0.8%.
3) Bubbles
Increased savings (and plausibly, decreased elasticity of savings to yields) and decreased need for investment cause the prices of investment assets to rise. They seem to create new zero-or-negative-yielding investment assets, by creating digital assets and by turning collectibles, art, fine wine, and so on into "asset classes". They create a need in the market for hype companies which promise some futuristic technology but cannot actually deliver it.
Arguably, a large enough demographic shift in the supply of and demand for investment opportunities changes the nature of the economy, and therefore the function of the banking system and central bank, from investment-driven to savings-driven. When we think of economic history, especially US economic history, we think of an economy with rapidly expanding credit and real assets, with many possible new investment opportunities clamoring for savings to be invested in them, leading to high yields on assets and many productive opportunities going unfunded. In turn this led to a credit system driven by a vigorous search for the best investment opportunities so as to maximize scarce savings, and a central bank whose purpose was to serve as lender of last resort if or when this vigorous search got overleveraged and had an accident. Such a central bank would not need to pay interest on its liabilities and would charge a high rate to lend, effectively setting a floor (zero) and ceiling (its intervention lending rate) on all good credit rates in the economy. But a savings-driven economy has stagnant or shrinking real assets, with few possible new investment opportunities, with low yields on assets and many unproductive opportunities getting funded. It has a credit system driven by a desperate search for any plausible investment opportunities so as to deploy excess capital, and a central bank whose purpose is to serve as borrower of last resort, manufacturing sufficiently high-yielding liabilities in which other actors can place excess funds, setting a floor (its borrowing rate) on all credit rates in the economy.
This excess of savings and shortage of new investment opportunities should massively increase the price of investment assets. Not only that, but it creates bubbles. As the unproductive zero-yielding assets created by this shift to a savings-driven economy rise in price, especially if they outperform actual yielding assets, they gain speculative momentum and may become rational to invest in in the short-run. And so it is that an enormous variety of non-yielding collectibles have risen enormously in price in recent decades. A Black Lotus from the nerd card game Magic The Gathering (even the 1993 Unlimited edition, the cheapest by far of the three official tournament-legal Black Lotus printings) sold for $1,057 in December 2010 and $17,050 in May 2026. The Rare Whisky Icon 100 price index of valuable whiskies rose from 319 at inception in 2016 to a peak of 923 in 2022 (and is since down to 550 in June 2026); the WatchCharts Rolex Market Index (for secondary market Rolex prices) similarly went from 16,000 at inception in 2017 to a peak of 41,000 in 2022 and is since down to 29,000 in June 2026. The 1995 Lego spaceship set Star Hawk II - which I received from my grandmother for my 9th birthday, one of my most beloved childhood gifts - in June 2026 now sells, if sealed and complete, for an estimated $488, up from an original RRP of $26.99. Glove Cases for the video game Counter Strike 2 sold for $0.01 throughout 2018, and on June 7 2026 the best bid is $23.09; the Chroma Case for the same game traded from $0.04 to $0.09 throughout 2018 and now have a best bid of $6.99; Operation Hydra Cases reached a peak price of $3.96 in 2018 and right now sell for $50.90. There are endless more examples, from handbags through classic cars and so on, but the best examples are cryptocurrencies. They never achieved widespread use as an alternate means of exchange - except for in crime, which is indeed a big market - but as their prices rose a significant industry built up around them to convince people to invest in them, to develop new cryptocurrencies, and so on, which allowed them to achieve some use purely as a (mostly fictitious) store of wealth.
These also illustrate most clearly how these bubbles can draw talent and initiative away from other potentially productive activities. Many of the most skilled software developers of the 2010s went to work in crypto, and it was rational for them to do so, and they were rewarded highly for it. And, today, even among crypto fans the crypto world is known to have far more than its fair share of grifters, scammers, and deluded zealots, because bubbles driven by an excess of demand to save over real investment opportunities create a profitable market niche for these characters. If extended, such bubbles encourage more people to adopt more aspects of grifter and scammer into their own personas and ultimately, if extended for long enough, presumably increase the frequency of genetic alleles linked with such behavior. While not as prominent and widespread as in crypto, there are plenty of examples of such characters in public market equities (and perhaps in politics) too. Negative population growth driving bubbles and therefore grift is just one of many ways in which negative growth incentivizes negative-sum behavior on the part of actors in the system.
Many researchers have considered the likely effects of demographic change on economic variables and asset prices - Arnott and Chavez, Juselius and Takats, Goodhart and Pradhan - but to crudely summarize their work, the simplest (and therefore likely most accurate) model of demographic effects is that higher dependent shares cause higher inflation and lower asset prices, while lower dependent shares cause lower inflation and higher asset prices. Because low dependent shares mean a more productive economy (as more of the population is able to participate in the economy) and more savings potential, while high dependent shares require dissaving to support them and mean less of the population is able to contribute.
And some part of the explanation for low inflation and high asset valuations, on average, in recent decades is that dependent shares fell to historically low levels and are now only rising slowly. In fact, in most countries with low birth rates, they will rise very slowly and even with a rising elderly proportion of the population will not rise to historically high levels.
The reason is: not just the elderly, but also children, are (expensive, economically burdensome) dependents. While a birth rate decline eventually inverts population pyramids and leads to an excess of elderly dependents, it immediately reduces the burden of young dependents. The aged economies of the future, in most reasonable projections, will not actually have a much higher dependent share than the young economies of the past. During China's demographic transition, from 1980 to the early 2020s, its total young and old (<20 and >65) share of the population fell from 50% to 35% while its birth rate fell from 2.7 children per woman to 1. Now that it has 1 child per woman - its total young and old share of the population is forecast to rise from 35% back to 50% by 2080. The US' total young and old dependent share is forecast to rise from 41% today to 46% in 2080; Japan's from 45% today to 50% in 2080. For comparison, the US' total young and old dependent share was 48% in 1960.
What this means is that we will be no (or only barely) more burdened by elderly care and elderly nappy-changing than we were in the past by childcare and baby nappy-changing - as long as we remain within the comforting spiral of demographic decline. It is only when or if birth rates rise again that total dependency ratios will get much higher than were seen in the past.
This may be the most important mechanism of all. Falling birth rates are, in the short run, economically very pleasant. They immediately reduce the dependency burden, freeing up labor and capital for other uses, driving up short-run growth. If they have negative effects these do not kick in for decades. It is easy to imagine both governments and market systems being subtly incentivized to bring them about. But rising birth rates are, in the short run, economically painful. Once polities have gotten used to low birth rates it is hard to imagine - as much as politicians might overtly claim to wish to increase birth rates - that the negative economic effects of a major increase in birth rates could be tolerated.
4) Political economy
The inversion of the population pyramid also drives a change in the political economy of asset prices. A population with a majority of young voters is indifferent to asset prices or prefers lower asset prices. A population with a majority of old voters prefers high asset prices. As voters get their way and asset prices rise, the resulting rise in wealth inequality exacerbates the extent to which effective political power speaks for asset owners. This ultimately leads not just to high housing (and bond, and equity, and gold) prices but to pro-incumbent policy: with with less enforcement of antitrust, monopoly and so on, regulations that stifle potential new competitors to existing incumbents, a weakening of the political power of labor, and ultimately much higher profit margins; with restrictions and delays on building new housing and real assets and infrastructure. In a democracy, an inverted population pyramid leads to more voting power for incumbent capital owners (who are likely to be old). It is in the interest of incumbent capital owners not to allow the creation of new capital anywhere the new capital could be competitive with theirs. This inability to create new real capital - houses and so on - prevents births, cementing low fertility and an inverted population pyramid.
According to US Census data (historically broken into age brackets 18-24, 25-44, 45-64, and 65+) the 1984 election had 53 million voters aged 44 or under, and 49 million voters aged 45 or older. In the 2024 election, there were 60 million voters aged 44 or under, and 95 million voters aged 45 or older. Youth participation has actually risen over that time (by 3.1% for the 18-24 year-old age group) but by less than elder participation (with the proportion of 65+ year olds voting rising by 4.0%.)
Older voters are more likely to own homes, and much more likely to own multiple homes. They are more likely to own equity, and much more likely to own lots of equity. In the extreme (65+ voters went from being 18% of the vote to 28% of the vote from 1984 to 2024), they are much more likely to derive all of their income or consumption from assets and not from labor. Of course, they also receive benefits which make up an increasingly large share of the economy and of federal tax receipts. Social Security plus Medicare went from 5.9% of US GDP in 1984 to 8.0% in 2024 - from 35% of US federal tax income to 47%. The government deficits this has driven have, in turn, been allowed by the global excess of savings resulting from ongoing demographic transition. The increasing share of the voter base receiving these transfers has become a powerful special interest group dedicated to these transfers. As these transfers become an increasing share of the government budget they reduce fiscal and political headroom for measures that could increase the birth rate.
The extent to which local governments (dominated by elderly participation) have blocked new housing development in key economic hubs has been widely remarked on. Across the US (according to Federal Reserve data series) residential property prices have risen 14.4 times since 1975, while commercial property prices have risen only 6 times. A large part, if not all, of that outperformance will be the rise in the value of the rent attached to the legal right for a property to be residential. Permitting and connecting energy generation, and new development of natural resources, have gotten harder too. It is possible that personality changes associated with ageing - rising risk aversion, or rising aversion to change - may directly motivate some of this as well as economic incentives. This limitation on actual physical growth - not measured economic growth, but growth in actual energy, materials, living space - obviously inhibits birth rates.
Meanwhile, increasing concentration of capital, and perhaps the increasing effectiveness of mass public manipulation technologies in the information age, have led to many political systems, especially the US, moving generally closer to one dollar one vote than one man one vote. (In the US this is exemplified by Citizens United vs FEC, but actual election spending has risen a surprisingly small amount in the past few decades.) Ultimately, it seems, while individual firms compete with each other for labor and for consumers, in aggregate the negotiation between capital and labor over the division between the fruits of their joint product is - a pure political struggle. And in monopolistic markets so too is the rent that capital can take from consumers. While in the past the median voter expected to have more future labor income than they then had capital, today the median voter has more capital than labor income. So the median voter now benefits when corporations take more than what had previously been their fair share from labor and from consumers. And so the government has tacitly encouraged profit margins to rise to historically unprecedented levels (after-tax profits were 9.2% of gross domestic income in 2024, above the pre-2020 record of 8.9% set in 1929, and far above the 1945-2000 average of 5.3%), justifying the price increase in US equities that was driven by the savings / investment imbalance, further entrenching the wealth and thus political power of US equity owners. To some extent this suppression of wages in favor of return on capital - which in aggregate transfers income from young to old - inhibits birth rates over time, but it may support birth rates in the short run while the bubble is rapidly expanding. South Korea (with the lowest birth rate of any major country in the world) saw its birth rate rise 16% from the first quarter of 2025 to the first quarter of 2026 - after a year in which its equity market rose 93% in dollar terms.
While all the mechanisms described above "make sense" - should clearly work, in isolation - their effects are refracted through the endless complexity of the economy. While the demographic transition is happening across most of the world - all but the very least developed countries, not just advanced economies but all economies with effective trade and cultural links to advanced economies - excess savings are more pronounced in some countries than others and different countries have different relationships with the imbalances caused by the demographic transition. In practice, in the 2020s, excess savings are above all concentrated in China, whose government forces its people to save in this manner and attempts to direct those savings into continuous massive capital expansion. A small portion of those savings leak out and overwhelm already tight markets in the rest of the world's supply of investment opportunities. Simultaneously China keeps its currency significantly undervalued and uses its rapidly expanding real capital base to dump on export markets globally.
Excess investment is above all concentrated in the US, which - as the world's reserve currency, largest consumer market, and cultural hegemon - is naturally positioned to host the new natural monopolies (search, social networking, spreadsheets) that arose with the information age. Whether LLM-type AI will turn out to be a natural monopoly too remains to be seen. As US assets have been steadily pushed up by this imbalance, they have attracted momentum flows and gone into a bubble. As a result the US has become generally uncompetitive, except for a few superstar industries able to use the high price of US assets to attract the world's most ambitious people.
A number of other countries, mostly China's neighbors, have followed China in having a low-rate, low currency, financially repressive policy. While it makes sense for countries facing most rapid demographic decline to encourage their residents to accumulate foreign assets, we can perhaps unfairly stereotype the situation as: the residents of these countries work extremely hard, never seeing the fruits of their labor, to provide us with most of our material goods and to transfer vast profits to US corporations, in exchange for tokens we conjured out of air, and any disruption to this state of affairs will lead to us devaluing their hard-earned tokens. On the other hand, it has allowed them to become more advanced in most forms of technology and manufacturing, while the US suffers from a terrible form of Dutch disease, sitting as it does on an infinite supply of the most valuable natural resource in the world, reserve currency status.
5) Culture
So in aggregate, the economic imbalances induced by below-replacement birth rates drive fertility further down - ultimately, in essence, because driving fertility further down is the easiest way to prevent already-high dependency ratios from worsening further. But low fertility reinforces itself culturally as well as economically. Above all, humans are mimetic and easily influenced by social norms. Marriages and childbirths appear to be contagious within peer groups. To be normal is, always, aspirational. So in societies with high birth rates people will naturally learn from childhood that high birth rates are normal, healthy, and good. They learn this from their families, friends, and communities more than from media - although today as humans are less strongly socially connected, and have increasingly strong parasocial relationships via media, media may be more effective than it was in the past. Today, while children are (generally) viewed as aspirational, larger families are increasingly uncommon and much later age of marriage has become normalized or favored in advanced economies, and this must lower birth rates somewhat in the short run (as it spreads births out into the future) and slightly in the long run (as the risk of failing to marry or of later medical, etc., problems preventing children increases).
Of course, fundamentally, as humans in a physical world sharing various kinds of real infrastructure, as fewer people have children the cost of having children rises subtly in various ways. There are fewer daycares around, people judge you more when the baby cries on a plane, you have to turn down more social invitations as they would be unwelcoming to children. That kind of thing. But the cost of having children has risen in various more direct and hard-to-understand ways which may be linked to the fertility feedback mechanisms described here. Education and childcare have become enormously more expensive, rising much faster than inflation in most advanced countries (this may be purely because of the Baumol effect). Harder to quantify but universally acknowledged, the time and effort cost of children has risen significantly, especially in the US. Upper-class children require far more time-(and money-)intensive activities to culture them sufficiently to compete in their admissions arms races; all children now require permanent carers, where once they could be left to fend for themselves. Parents who let children venture out or play on their own are now considered willfully negligent and are at risk of serious consequences. This adds multiple hours per day to the time cost of children - leading to the plausible argument that children have now become a superior leisure good. This stigmatization of unsupervised freedom for children quite likely makes them less well equipped for adult life than they would have been otherwise.
Rising safety concern for children drives laws that directly increase the price of children as well as increases their time and attention cost. The famous example is child car safety seat legislation, which economists determined drove a 0.7% reduction in birth rates for families with two existing children - who would thus likely have to upgrade to a bigger car if needing a child safety seat - compared to similar families in states without child safety seat laws. But examples will abound: window guards, daycare regulations, playground regulations... These rules surely do marginally make children safer. But many of them will have a larger marginal impact on birth rates.
In short, as birth rates have fallen, children have become more and more valuable and thus more and more expensive. And cultural standards have shifted towards a low-fertility norm. It is possible to imagine these trends being reversed by determined policy effort but, for now, easier to imagine them continuing.
Conclusion
Low birth rates cause low birth rates. They do this economically, politically, and culturally. They cause economic and social pressures that raise the price (in money and time) of children, reduce living space and real resources available for children, and ultimately lead towards an economic state in which the short-run dissaving or dependency pressure of additional kids becomes intolerable.
Enormously more could be said about many factors which contribute to fertility rates - some of which, through the complexity of the economy, may actually bear somewhat on the mechanisms described here and their interactions. But the argument over what has caused the fertility decline is interminable and there is no-one left to convince, all commentators have already picked their sides. This is my best brief effort to explain why low fertility matters: because low fertility begets low fertility, and thus becomes a trap that will take a remarkable shock or incredible policy effort to break out of.