LONDON — While we should always be wary of proposing a universal law of human nature, the past half-century has revealed a consistent and enduring pattern. In every society that combines economic prosperity, a well-educated female population, and access to safe and affordable contraception, fertility rates fall below – often well below – the replacement level of 2.1 children per woman.

 

Sweden’s fertility rate dropped below 2.1 as early as 1968. By 1975, most European countries, along with the United States and Canada, had followed suit. None has risen above 2.1 since then.

 

As women’s education and access to contraception expanded, low fertility became increasingly common. India’s fertility rate now stands at 1.96; Latin America and the Caribbean fell below replacement level in 2015 and now stands at 1.8; Vietnam, Malaysia, and Turkey have rates of 1.9, 1.54, and 1.62, respectively.

 

A similar pattern appears within countries. In India, the fertility rates of the three most economically developed states – Gujarat, Maharashtra, and Karnataka – range from 1.7 to 1.9, whereas rates in the poorer states of Uttar Pradesh and Bihar are 2.4 and 3.0, respectively.

 

Over the past 50 years, the world’s fastest-growing economies – South Korea, China, Taiwan, and Singapore – have recorded the lowest fertility rates, ranging between 0.8 and 1.2 children per woman. By contrast, the highest fertility rates are found where poverty remains entrenched or women’s rights are severely restricted, such as Sub-Saharan Africa (4.26) and Afghanistan (4.76).

 

Yet, instead of being welcomed as the natural outcome of economic progress and women’s freedom to choose, declining fertility is often framed as a looming crisis. In India, commentators warn of the “end of the demographic dividend”, while low fertility rates in Europe (1.4) and North America (1.6) are routinely portrayed as evidence of “stagnant” societies suffering from demographic “exhaustion”.

 

In 2018, the Robert Schuman Foundation warned that Europe risks “demographic suicide”. In the US, billionaire Elon Musk has predicted “catastrophic population collapse,” and Vice President J.D. Vance says he wants “more babies in the United States of America”.

 

But this conventional wisdom is deeply misguided. While East Asia’s exceptionally low fertility rates will indeed create serious challenges if sustained indefinitely, fertility rates in the 1.5-2 range are likely more conducive to human welfare than those above the replacement threshold of 2.1.

 

Moreover, the greatest demographic challenge of our time is not falling fertility in successful countries. Instead, that distinction belongs to explosive population growth in many of the world’s poorest economies, particularly across Africa.

 

Too few workers?

 

Warnings about falling fertility often focus on the old-age dependency ratio – the number of retirees relative to the economically active population. In countries where fertility decline started earliest, McKinsey estimates that the ratio of people over 65 to working-age individuals (aged 15-64) will rise from around one-quarter today to one-half by 2050.

 

There are two major problems with this widely used measure. First, it assumes that retirement ages remain fixed despite rising life expectancy, thereby conflating two distinct demographic trends: falling fertility and increasing longevity. As the United Kingdom’s Pensions Commission (which I chaired from 2003 to 2006) showed, a significant share of the apparent rise in dependency – about half in Britain’s case – disappears if retirement ages are gradually raised so that the proportions of adult life spent working and in retirement remain roughly stable.

 

Second, the standard measure ignores the fact that children are also dependents, which means lower fertility reduces child dependency even as old-age dependency rises. Politicians calling for “more babies” should recognize that if birth rates do increase, the total dependency ratio will rise even faster than before until those children enter the workforce two decades later.

 

But, in principle, the dependency ratio argument is valid: lower fertility means a one-off increase in the number of retirees relative to workers. The crucial question is whether productivity growth can make up the difference. So far it always has.

 

In 1800, old-age dependency ratios were close to zero, because most people worked from childhood until death. In today’s advanced economies, there are about 0.4 retirees for every worker, each of whom enters the labour force far later in life and works far fewer hours each year than their counterparts in the early 19th century. The share of hours lived after the age of 15 which are spent working has fallen at least 60 per cent; but advanced-economy GDP per capita has increased 15-fold.

 

This reflects humanity’s extraordinary capacity to raise productivity. In pre-industrial societies, most people worked in agriculture, laboring more than 60 hours per week simply to produce enough food to survive. In developed economies today, fewer than 3% of working-age adults – working far fewer hours – produce enough food to feed not only themselves but also retirees, children, and teenagers, who no longer need to work.

 

The same is true of manufacturing, which currently accounts for less than 20 per cent of the workforce in developed economies. China’s 125 million factory workers produce 30 per cent of all the manufactured goods used by the world’s 8.2 billion people; and robot-based automation will reduce that number quickly even as output continues to rise. While the technological advances of the past 50 years have given us mobile devices with far more computing power than NASA used to land a man on the moon, producing all the world’s smartphones, laptops, and tablets employs only 10-15 million people out of a global working-age population of five billion.

 

So, rising dependency ratios are not a problem unless humanity has suddenly lost its ability to sustain productivity growth. In fact, AI is likely to strengthen it.

 

The AI acceleration

 

At first glance, AI may appear just another tool for automating an ever-wider range of tasks, but its self-learning capacity makes it more than that – not just a technology which can increase productivity, but one which can itself accelerate technological advance. This transformative potential underpins optimistic projections of AI’s economic impact, with true believers predicting that it could automate almost all jobs, boosting annual growth to 20-30 per cent, compared with the twentieth-century average of 2.8 per cent.

 

There are good reasons to take such extreme predictions with a very large grain of salt, especially those suggesting that rapid gains in underlying productivity will be reflected in measured GDP. But we are clearly moving toward a future in which machines can perform most human jobs. By 2100, all the world’s food could be produced by less than 1 per cent of the global population, and manufacturing, transport, and logistics could require a similarly small fraction of the workforce. Many people may remain employed in these sectors, but only because high fertility rates in poorer countries continue to generate surplus labor willing to work at very low wages, not because it is technologically necessary.

 

When it comes to desk jobs, humanity has shown an almost limitless capacity for inventing tasks devoted to zero-sum competition, especially in areas like marketing, sales, lobbying, and finance. But AI is still poised to automate a significant portion of the roughly one-third of jobs that involve repetitive tasks like information gathering and processing.

 

Tasks that require hand-eye coordination, fine motor skills, and flexibility will be far harder to automate. Robots are still pretty inept at something as simple as loading a dishwasher; plumbers and electricians are unlikely to be replaced any time soon. But even here, the question is almost certainly when, not if.

 

Bank of America estimates that by 2060, there could be two billion humanoid robots working in people’s homes, alongside another billion in the service sector. If that forecast is even remotely accurate, and if these machines are even moderately capable, there will be no shortage of workers.

 

There are also jobs we should not automate, even if we could. In the US, about 14.5 per cent of employees work in health care and social assistance. But even in this sector, only a minority of hours worked involve emotionally important face-to-face interaction. In 2017, McKinsey estimated that 36 per cent of hours worked in this sector could be automated with already existing technologies. As that percentage rises over time, and as work is automated across all other sectors, finding enough workers will hardly be a major challenge.

 

For many people, a world of near-limitless automation raises another concern: Will there be enough jobs for everyone to earn a good living? Whatever the answer, it is absurd to worry about a shortage of jobs while simultaneously fearing that low fertility will leave us with too few workers.

 

Too few innovators?

 

While machines can perform many tasks, what about innovation and creativity? In a world of low fertility and population decline, “there are fewer young people around to think, create, and invent,” argues Paul Morland in his 2024 book “No One Left: Why the World Needs More Children”. Fewer births, the argument goes, means fewer discoveries.

 

But the notion that “old people” cannot innovate is ludicrous. Whatever one thinks of his politics, Musk at 54 shows no signs of declining ability to drive technological and business innovation. Beethoven wrote some of the most strikingly original works in music history when he was in his fifties, and Picasso was as inventive in his sixties and seventies as he was in his twenties.

 

Still, on average, a higher share of younger people may contribute to innovation, especially in technical fields where mathematical reasoning plays a central role. But only a fraction of the population, whether young or old, has ever been responsible for major technological advances.

 

Consider nuclear science. In 1890, humanity had almost no understanding of nuclear physics; by 1945, its knowledge was deep enough to build atomic bombs and generate nuclear power. Yet, the scientists and technologists behind these breakthroughs numbered only in the thousands and were concentrated mainly in Europe and the US at a time when these regions’ youth population was less than a tenth of today’s global total.

 

And that was before the development of AI, itself developed by a tiny fraction of the global workforce. DeepMind’s AlphaFold has mapped the structure of nearly all known proteins, hugely improving the ability of very small numbers of research scientists to develop new drugs.

 

The same applies to entertainment, fashion, and the culinary arts, where new ideas are driven by a small minority of creative people, many of whom make their most important contributions well past youth. And if we really did need more innovators, having smaller families later in life might free up youthful energy otherwise devoted to childcare.

 

South Korea is a case in point. Its fertility rate fell below 1.7 in 1985, and the share of its population aged 20-40 has declined by 20% since 2000, yet the country still ranked first in Bloomberg’s 2021 Innovation Index. Earlier this year, it ranked twelfth in the Brand Finance Global Soft Power Index 2025, a “meteoric ascent” attributed to its “dominance in arts and entertainment” and the worldwide appeal of its cultural exports, whether K-pop, K-dramas like Squid Game, or K-beauty products.

 

At just 0.8, South Korea’s fertility rate may eventually sap its innovative vitality. But the idea that low fertility rates inevitably result in technological and cultural stagnation is supported by neither logic nor empirical evidence.

 

Adair Turner, chair of the Energy Transitions Commission, was chair of the UK Financial Services Authority from 2008 to 2012. He is the author of many books, including “Between Debt and the Devil: Money, Credit, and Fixing Global Finance” (Princeton University Press, 2016).