In my last article, I outlined the unusual demographic future of humanity: it is highly probable that, between 2050 and 2060 our species will reach a population peak close to 9.5 billion, which will then begin to fall. I also pointed out some of the mechanisms that could make this prediction incorrect, such as a rise in average life expectancy, changes in the relative proportions of social groups, government policies, etc. As it happens, China has just announced new and very restrictive regulations on abortion, the purpose of which is to increase fertility.

Still, let us assume that the future approximately resembles the demographic trends we observe from our vantage point in 2021. It is worth spending some time thinking about the consequences of this population change. As is almost always the case, there will be positive and negative consequences, and which of these predominate will depend, in part, on our ability to design policies that manage the transition well.

In this essay, I will focus on the most direct economic consequences. In the third and last essay of this series (coming out in a few weeks), I will review the social consequences of this demographic shift on the structure of the family, and on the distribution of the population in terms of both land and housing.

Positive Consequences of Population Decline

Start your day with Public Discourse

Sign up and get our daily essays sent straight to your inbox.

The main positive consequence that comes to mind is less pressure on natural resources. A smaller or slower-growing population makes it easier to decarbonize the global economy and conserve areas of biological wealth, keeping them from being affected by human activity. This consequence reinforces another key transformation of the global economy: the gross domestic product of advanced economies is becoming progressively “weightless,” and therefore requires fewer natural resources to generate.

Let’s think about a simple example. I currently have 1,833 books on my Kindle. Each shelf in my office can hold, on average, about forty books (some bigger, some smaller). In other words, I would need about forty-six shelves to store the books on my Kindle in physical format. That would take up my entire office and a good part of another one. Since I started reading and buying books as a child, until 2010 when I switched to e-books, I progressively accumulated “weight” in paper. From 2010 until today, my paper “weight” first stabilized and then later began to fall. In 2020, I bought only fourteen paper books, but I gave away more than fourteen of the books I already had. There were many books that I had repurchased in electronic version to have them always at hand, so I gave the physical copies to my students. In other words, my book collection, being much larger on January 1, 2021 than it was on January 1, 2020, still “weighed” much less. This observation applies to almost all my possessions: I can’t even remember the last time I bought a CD or a DVD. All my possessions together “weigh” less today than they did five years ago.

The gross domestic product of advanced economies is becoming progressively “weightless,” and therefore requires fewer natural resources to generate.


Modern economic growth does not, in general, use more resources per capita. We get better and better at recombining ideas in incredibly creative ways to generate much more added value. That’s why the UK was able to emit less carbon dioxide in 2019 (before the pandemic) than it did in 1890, despite having a gross domestic product, in real value, thirteen times larger. For its part, the United States consumed 2 percent less oil in 2019 than in 1978, while its gross domestic product was, in real value, three times larger (and the United States has not been overly concerned with saving energy).

The weakness of the above argument is that it focuses on more advanced economies. Less developed economies still need more resources to grow and reach the level of wealth of the leading countries. But as these economies stabilize in total size, as a result of population decline, even higher per capita resource use will be compatible with stable or declining global resource consumption.

The Negative Consequences

The negative consequences of the fall in world population are more complex. Most importantly, we must get used to the fact that gross domestic product is growing at lower rates, and we must adapt our societies accordingly. Historically, labor productivity has grown by an average of 2 percent per year. If the working population increases at 1 percent per year (as it did during the 1960s in many advanced economies), in a normal business cycle, the economy will grow by 3 percent (2 percent productivity plus 1 percent population). When the economy accelerates above the normal situation (due to demand or supply shocks), it grows at 4 percent or 5 percent. When it slows down, it grows at 1 or 2 percent. Overall, though, it is always fluctuating somewhere around the 3 percent mark.

Now, let’s imagine a situation where the working population falls at 1 percent per year instead of growing at that rate. If labor productivity continues to grow at 2 percent—I’ll come back to this in a moment—the economy will grow on average by 1 percent (2 percent productivity minus 1 percent population decline), not at 3 percent, as it did before. When the economy accelerates above average, it will grow at 2 or 3 percent. When the economy slows, growth will be at -1 or 0 percent.

This is exactly what has happened to Japan. From the outside, its economy has looked stagnant since the mid-1990s. But if we look at it in terms of its gross domestic product divided by working-age adults (ages sixteen to sixty-five), a measure of the economy’s potential workers regardless of whether they are employed or not, Japan has grown at about the same rate as the United States and faster than Germany. The rivers of ink written about the origins of Japan’s “economic ills” over the past few decades are basically useless. It’s not the fault of the Bank of Japan, or China, or anyone else. Japan is growing at the speed one would expect given its demographic changes.

You might be asking: what does it matter that the economy is growing at 1 percent instead of 3 percent in total terms, if in terms of working-age adults we continue to grow at the same rate? Shouldn’t we be concerned about per capita income instead of total income? Yes and no. Yes, because the relevant measure of income to assess the welfare of a society is per capita, but no, because per capita income is not everything. The ability to pay our public debt, for example, depends on total income, not per capita income.

Imagine a country with a public debt to gross domestic product ratio of 100 percent. If that country’s economy grows at 3 percent, public debt stabilizes as a percentage of gross domestic product. If the general government deficit is 3 percent, we add 3 more points of debt to the numerator, but the denominator also grows by 3 points. On the other hand, if the economy grows at 1 percent, we need to bring the general government deficit to 1 percent to stabilize the debt. This takes considerably more fiscal effort.

The same is true of many state obligations, such as pensions and public health care. These social benefits are much more onerous to maintain when, along with the fall in the working-age population, we are faced with a lengthening of life expectancy. While social costs rise rapidly, the economy lags behind. Changes such as raising the retirement age help, but they do not significantly change the basic scenario. By working longer, we can slow down the fall in the active population, but we do not stop it.

In fact, the arguments in the previous paragraphs may be too optimistic. Will labor productivity continue to grow by 2 percent even if the population falls? There are reasons to doubt it.

Knowledge, Creativity, and Innovation

Earlier, I stressed that economic growth is based on recombining new ideas. What do I mean here by ideas? I mean any creation of the human mind that allows us to transform a resource into a good or service. For example, one idea many centuries ago was to use silicon to make concrete. More recently, it occurred to us to use silicon for transistors (like the ones you are using to read these lines). Just a few weeks ago, there were major breakthroughs in using silicon to make better batteries for electric cars. But other ideas include new theorems in mathematics (important, for example, for your credit card to work on the internet) and the development of new forms of business. Founding Amazon or the invention of cost accounting were as much innovations as a new chemical process.

And who creates these ideas? People. Ideas are the result of research effort and creativity. Therefore, all else being equal, the flow of ideas in a society depends on the number of people thinking. It is virtually certain that a society with 200 million people will develop more innovative ideas than a society with two million, provided that the economic institutions are similar. Switzerland produces more ideas than Nigeria because it has better institutions, but Switzerland produces fewer innovative ideas than the United States because it has fewer people. Smaller populations, even in the countries of the world with better innovative institutions (United States, East Asia, Western Europe), are going to be populations with fewer innovative ideas.

Who creates ideas? People. Ideas are the result of research effort and creativity. Therefore, all else being equal, the flow of ideas in a society depends on the number of people thinking.


This phenomenon may be particularly pressing if we note that the evidence suggests that, as we accumulate knowledge, it is increasingly difficult to create new ideas. Humanity will never have a new Galileo Galilei, Isaac Newton, or Charles Darwin. The foundational ideas in physics, mathematics, and biology that these scientists developed cannot be “reinvented.” Even the relativity theory or quantum mechanics of twentieth-century physics, with all their novelty and dramatic consequences, were more refinements of existing cognitive structures than radical breakthroughs, as was the invention of the experimental method or the development of infinitesimal calculus. Newton was probably not wrong when he argued that he could see further because he stood on the shoulders of giants, but even from up there more and more clouds prevent us from having a clear view.

I notice it a lot in my daily work. Soon, I will start teaching a PhD course on optimal control. This is the field of applied mathematics that analyzes how to control a dynamic system to achieve the best possible results over time. In economics, for example, this theory is applied to study how to manage a portfolio of investments over time to maximize its return, given a certain level of risk. When I was a student in the late nineties, I estimate that it took about twelve months to specialize in the field (once you finished the basic PhD courses) and to be ready to start writing a groundbreaking thesis at a global level. Today, I estimate that it takes at least twenty-four months. The number of things we have learned about optimal control in the last two decades is spectacular. But this also means that coming up with new ideas in this field is more complex than ever, and one has to learn many more things. The fathers of this specialization, Richard Bellman and Lev Pontriaguin, had it much easier: everything was yet to be discovered. This does not detract in any way from their achievements (they were both great geniuses), but it does put the merits of the current generation of young researchers into perspective.

For the last three centuries, humanity has been participating in a race in which, on the one hand, it is increasingly difficult to come up with new ideas, but on the other hand, there are more and more of us engaged in research. These two forces have counteracted each other, and the result of the race, so far, is that we have been able to develop new technologies and scientific advances that have given us that 2 percent average annual growth in labor productivity.

With a falling population, we will start to lose the race. It will be increasingly difficult to increase the total number of academic researchers and business innovators. Many people have no interest or ability to do research, and each additional researcher is one less person working in the production of goods and services. As a society, we can’t all dedicate ourselves to innovation. Someone will have to apply that innovation to produce things we like.

In other words, a shrinking human population can also be a much less dynamic population from the point of view of innovation. Many economists argue that the effects are already being felt in the creation of new businesses and in the poor productivity since 2008.

To return to our calculations: let’s assume that, as a result of decreasing innovation, labor productivity will only grow by 1 percent per year. With a 1 percent fall in the working-age population, the “new normal” will be 0 percent growth in gross domestic product. How are we going to pay for the welfare state and our level of public debt with zero average growth? It doesn’t add up.

With a falling population, we will start to lose the race. It will be increasingly difficult to increase the total number of academic researchers and business innovators.


Robots Won’t Save Us

I often come across the argument that automation will save us. If we have enough robots and develop artificial intelligence systems, the argument goes, we can continue to grow without a problem. My response is always that this was the idea behind the Soviet Union’s economic strategy; it didn’t work in the twentieth century, and it won’t work now.

In 1928, Stalin embarked on a program of accelerated industrialization based on very high rates of investment. The idea was to accumulate enough physical capital to be able to produce enough consumer goods in a few decades to achieve communism. The problem with this investment strategy is that the accumulation of physical capital is subject to diminishing marginal returns. Building the first blast furnace in Magnitogorsk in 1932 had very high returns: we go from producing no steel at all to producing steel that has a very high marginal value. The second blast furnace in the 1950s had a much lower yield. The last blast furnace, completed in 1987, was almost worthless. Think about it with the televisions in your house. The first television we buy, which we put in the living room, is very useful. The second television, for the bedroom, is fine, but it’s no longer the same. The third television in the kitchen helps, but not much. The fourth television stays in the cupboard. Just by looking at the diminishing marginal returns of physical capital installed in the Soviet Union one can understand, almost exactly, the high Soviet growth rates in the 1930s, the average rates in the 1950s and 1960s, and the stagnation of the 1970s and 1980s.

Robots are another form of physical capital, perhaps more interesting than a blast furnace, but not essentially very different. And as such, they are also subject to diminishing marginal returns. The first robot has high marginal returns, the second somewhat less, the third is of little use.

We grow because we have more and better ideas. We accumulate capital to produce ideas, not the other way around.


That is why I never use the word “capitalism” to refer to the economic system we have. Emphasizing the word “capital” makes us believe that the key to modern economic growth is to accumulate physical capital, be it blast furnaces or robots. No, it is not at all. We grow because we have more and better ideas. We accumulate capital to produce those ideas, not the other way around. Britain, and then the rest of the West, led the industrial revolution and modern economic growth because they were the first societies that, around 1650, began to have institutional structures that encouraged the systematic and constant production of new ideas. In short: with a smaller population in 2100, we are likely to have fewer innovative ideas and with them, less productivity growth.

Could I be wrong? Of course! Advances in deep learning in recent years may make it easier to develop new ideas even with fewer researchers. Ten years ago, I would not have ventured to guess that today we would have deep neural networks that can predict protein folding. Returning to the example of optimal control that I mentioned earlier, deep neural networks make it possible to solve problems that were impossible even two years ago. It is also possible that new technologies are not subject to the same diminishing marginal returns as the technologies of past centuries. For example, data are non-rivalrous goods: using data in one line of business of a company does not limit its uses in another line of business. By comparison, using a machine in one line of business does preclude its use in another line of business.

Will these changes be powerful enough to defeat the effect of population decline on the creation of new ideas? I, following Robert Gordon, tend to be pessimistic that these changes will be enough, but I would like nothing better than to be wrong.

This article was originally published in October 2021 at El Confidencial, a leading Spanish digital newspaper. We are grateful to Professor Jesús Fernández-Villaverde for his permission to publish it in English here, and to Thomas Howes for his translation.