SpeakFreely spoke with Steve Davies, Senior Education Fellow at the Institute of Economic Affairs, about his book The Apocalypse Next, global catastrophic risks, solar storms, nuclear war, supervolcanoes, and the difficult question of how much societies should spend preparing for disasters that are unlikely but potentially civilisation-ending.
SpeakFreely: Steve, thank you for being here.
Steve Davies: My pleasure. Thank you very much for having me.
SpeakFreely: Today we’re talking about The Apocalypse Next. One of the things you discuss is global catastrophic risk, and you distinguish those risks from smaller disasters. COVID-19, for example, was an inspiration for the book, but you wouldn’t classify the pandemic itself as a global catastrophic risk.
SD: That’s right. The distinction isn’t original to me. It comes from the work of the Oxford philosopher Nick Bostrom, who identified this category of global catastrophic risks some years ago.
There are two parts to the term: global and catastrophic. To qualify as a global catastrophic risk, something first has to be genuinely global. It has to affect the entire planet. A nuclear war confined to the Indian subcontinent—for example, a war between Pakistan and India—would be catastrophic for the region, but it wouldn’t necessarily be global enough in its effects to qualify.
Secondly, the event has to be catastrophic. Bostrom distinguishes between costs that are terrible but ultimately endurable and those that are genuinely unendurable. A catastrophic event is one that either results in human extinction or in the complete and irretrievable collapse of advanced civilisation. The crucial word there is irretrievable. So a global catastrophic risk is an event that affects the entire world and produces consequences that are catastrophic in that very strong sense.
SpeakFreely: One example you discuss is a Carrington event. What exactly is that?
SD: A Carrington event is a very large coronal mass ejection. The Sun regularly throws out large quantities of highly charged plasma from its atmosphere, the corona. That happens all the time, and every so often one of those eruptions is directed towards Earth. When a coronal mass ejection hits the Earth’s atmosphere, one of the effects can be the northern lights and other spectacular displays near the poles.
But occasionally you get a solar storm—a CME—that is exceptionally large. One such event was the Carrington Event in the middle of the nineteenth century, named after the astronomer Richard Carrington. At the time, it produced spectacular auroras but relatively little disruption to the world. What it did do was burn out telegraph wires in various places.
That is because when a really large amount of coronal material strikes the Earth’s atmosphere, it doesn’t just produce the northern lights. It can create electromagnetic effects capable of burning out electrical circuits, wrecking automatic controls and damaging computers. When the Carrington Event happened in the nineteenth century, there simply wasn’t much electrical infrastructure to destroy. Today would be a completely different story.
An official US government estimate suggested that if an event of comparable size happened now, it could reduce global GDP by somewhere between 10 and 20 percent. That is a catastrophic economic hit. To put it in perspective, that would be significantly larger than the damage to global GDP associated with the two world wars of the twentieth century. And the effects could last much longer.
It could crash power grids around the world, destroy large numbers of electrical appliances and systems, and seriously disrupt the internet. It’s a good example of something that was not a catastrophic risk in the past but could be one today. Paradoxically, our more advanced technology makes us more vulnerable to certain natural events because the disruption they would cause is so much greater than it would have been a century or two ago, let alone in pre-industrial history.
A Carrington-style solar flare is what we call roughly a hundred-year event. That doesn’t mean it happens neatly once every hundred years. It means there is approximately a one-in-a-hundred chance in any given year. We came quite close in the 2010s, when an eruption of solar material larger than the Carrington Event missed the Earth by a relatively small distance in interplanetary terms. And there have been periods in the past—for example during elevated solar activity many centuries ago—when much larger solar eruptions appear to have occurred.
So this is a potential event that could happen at any time. Fortunately, it is also a risk we can do something about. And that raises the next question: should we spend resources protecting ourselves against it, and if so, how much?
SpeakFreely: Given the potential economic damage, it sounds as though protection would be worthwhile.
SD: That’s the key point. Most global catastrophic risks are low-probability events. In statistical language, they’re tail events. But low probability doesn’t mean you should ignore them.
If the event is genuinely catastrophic, the damage is civilisation-ending. The cost can therefore be thought of not simply in trillions of dollars, but in the loss of all the future production, creativity and human lives that would otherwise have existed. Take full-scale nuclear war. That would clearly be global, and it could be catastrophic in the strongest sense because it could completely destroy advanced human civilisation. And for various reasons, it may be extremely difficult to recreate a technologically advanced civilisation afterwards.
So when you estimate the cost, you don’t just count the people killed or the immediate physical destruction. You also have to include all the future production, creativity and human lives that would have existed had the catastrophe not occurred. Once the possible loss becomes that enormous, even a probability of one in a thousand or one in a million becomes something you have to take seriously. In insurance, one way of thinking about this is value at risk: you consider the cost if the event occurs alongside the probability of its occurring.
If the damage is extraordinarily large, then even with a very low probability you can still end up with an enormous expected loss. So in many cases it makes sense to take steps to reduce the danger. Usually you cannot eliminate a global catastrophic risk entirely. What you can often do is mitigate its consequences. Then economics comes into the picture.
How many resources is it sensible to devote to protection? The higher the probability of the event, the more resources it makes sense to spend mitigating it. With something such as a Carrington Event, the risk is comparatively high. A one-percent annual probability is actually quite alarming given the potential cost.
One thing you could do is harden and stabilise electrical grids. You could require important electrical systems and infrastructure to be protected against electromagnetic pulses. There are other reasons to do that anyway, because EMPs could potentially be used in warfare. And if you spread the cost of hardening infrastructure over a sufficiently long period, it becomes a relatively sensible investment.
SpeakFreely: But there are other catastrophic risks where either very little can be done, or nothing can be done at all.
SD: Exactly. Suppose a sufficiently nearby star went supernova and the Solar System was hit by a huge burst of radiation. There is essentially nothing we could do about that. We would literally be toast. Fortunately, the probability is extraordinarily low. Then there are things such as supervolcanic eruptions.
Those occur at intervals throughout the Earth’s geological history and have happened during human prehistory. There was a massive eruption at Lake Toba in Sumatra which had an enormous effect on the global climate. Events of that scale are extremely rare—perhaps once every tens of thousands of years. So what is it worth spending to protect against something like that?
Probably not a huge amount. But it may still be worth doing something. A relatively inexpensive response would be what I describe as a kind of lifeboat measure. You could maintain resources such as the Svalbard seed vault: stocks of seeds, information, technology, samples of important crops and perhaps other biological resources. You could distribute these secure repositories around the world.
Then, if civilisation did collapse, survivors would at least have access to resources that could help them rebuild. That sort of protection is comparatively inexpensive, so even for a very low-probability catastrophe it can make sense.
SpeakFreely: So, in probability terms, a low-probability long-term risk can effectively become comparable to a shorter-term risk with a much higher probability if the potential consequences are large enough.
SD: Exactly. That’s how it works.
SpeakFreely: The seed-bank example is interesting because that seems like something that could be done voluntarily. You don’t necessarily need government to preserve seeds. But something such as requiring the electricity grid to be hardened would require government action. Does that create a risk of unintended consequences?
SD: It could. Let me pull back for a moment. Things such as the Svalbard seed vault are basically private action. There are lots of measures that I call “lifeboat measures” in the book that can be carried out through private action. They don’t require government. Other measures can be organised through voluntary collective action because they don’t require everybody to participate simultaneously. But there are some risks where you face a very severe collective-action problem.
Protection from a major solar flare may require coordinated infrastructure changes. Preventing nuclear war is another example. If you want to abolish nuclear weapons, it does you very little good unless every nuclear power gives them up. In cases like those, you are probably going to need some kind of state response.
But there is another distinction. Sometimes it may be enough for one or two sufficiently large states to protect themselves. Take a solar flare. You probably don’t need every government on Earth to harden its infrastructure in order to prevent an irreversible collapse of civilisation. If, say, the United States and China protected themselves effectively, that might be enough to preserve advanced civilisation even if other parts of the world suffered catastrophic damage.
But there are other risks—super-powerful AI, an abrupt and severe climate shift, or a pandemic—where protection may require genuinely global action. That is where things become much more difficult. You run into extremely serious collective-action problems because you need governments across the world to be broadly on the same page.
SpeakFreely: Suppose governments do decide to act against a global catastrophic risk. What kinds of unintended consequences should we worry about?
SD: There are two obvious broad categories. The first is that the whole thing becomes a job-creation boondoggle. It becomes simply another way for politicians and the political class to shovel rents into favoured client groups. That’s a risk with any government activity, including things most people regard as legitimate, such as defence spending, which in one sense is an enormous enrichment programme for private defence contractors and others.
You simply have to accept that risk and do what you can to guard against it. The other risk is more serious: that the measures you take impose severe costs on civil liberties or obstruct innovation more broadly. That is something you have to think about extremely carefully, because there may be cases where the cost of the protection is far too great relative to the likelihood of the event actually occurring.
The classic example is artificial intelligence. A number of people, including people who have worked directly in AI, believe there is a significant risk that artificial intelligence will escape human control. Some, such as Eliezer Yudkowsky, regard that outcome as effectively inevitable unless very strong precautions are taken. The fear is that AI could move beyond human control, achieve superhuman intellectual capabilities and potentially become sentient.
If that happened, the argument goes, there could be a serious risk of human extinction. Perhaps the AI decides that human beings are simply a nuisance. Or perhaps its objective is to keep the planet functioning well and it concludes that eliminating most humans is the best way of achieving that.
SpeakFreely: Essentially Terminator.
SD: Yes, exactly. According to people who take this view, there is therefore a non-negligible risk. There have even been extremely high estimates of the probability of AI contributing to human extinction.
What strikes me as bizarre is when people deeply involved in developing the technology themselves make claims of that kind. If you genuinely think there is an enormous chance that what you’re working on will wipe out humanity, the obvious question is: what on earth are you doing? The response tends to expose the collective-action problem: “If I don’t do it, somebody else will, so it’s better that I do it.” I’m not sure that follows.
But the broader point is that many people regard this as a substantial risk. People such as Nick Bostrom have therefore argued for extremely stringent global controls on scientific research, technological innovation and AI development. There are two problems with that.
First, you would need all the major powers to cooperate. That is extraordinarily unlikely because they face a prisoner’s dilemma. The enormous strategic advantage would go to whichever country allowed everyone else to restrict themselves while secretly continuing the research.
The Chinese, American and Indian governments are all perfectly capable of understanding that incentive.
But even if you leave that problem aside, there is a second danger.
Controls stringent enough to accomplish what is being proposed could require something approaching a totalitarian state.
They could also destroy innovation, which is one of the main mechanisms through which human well-being improves and, in some respects, one of the things that keeps our existing civilisation functioning.
So this could be a case where the cure—or the protection—is worse than the threat.
The attempt to avoid one catastrophic outcome could itself produce a catastrophic result.
That is why, when assessing measures designed to protect against global catastrophic risks, you have to include their downstream costs.
Those costs could be every bit as serious as the catastrophe you are trying to hedge against.
SpeakFreely: We talked earlier about Carrington Events and how a major solar storm would be much more damaging today because we live in a more technologically advanced and interconnected world.
Do you see globalisation as something of a Faustian bargain?
SD: It could be.
One of the things exposed by the COVID pandemic was the short-term fragility of globalised supply chains and just-in-time production and distribution systems.
Over the last fifty or sixty years, the world has in some respects become excessively complicated.
Production systems are highly distributed. They have been broken down into as many different stages as possible for reasons of economic efficiency.
That produces economic gains, but it also creates extraordinarily complex systems requiring enormous numbers of internal relationships, exchanges and contracts.
The problem is that systems like that can become very brittle.
A relatively small disruption in one part of the system can trigger what is called a cascading failure.
One part fails, which causes failures elsewhere, which trigger still more failures, until eventually the entire system collapses.
Catastrophic failures of complex systems have happened throughout human history.
You can even interpret the outbreak of the First World War in this way: as a cascading failure of the international-relations system that existed in Europe in 1914.
My concern is that the particular form globalisation and technological development have taken during the last forty or fifty years has created a series of systems that are excessively complex and therefore unusually susceptible to breakdown.
Oliver Letwin, the former Conservative MP and cabinet minister, wrote a book that I refer to about how completely dependent we have become on the internet.
Even a relatively brief interruption caused by something such as a solar flare or a major grid collapse could cause failures throughout the wider economy.
Food distribution could collapse.
Large parts of industrial manufacturing could stop because they depend upon just-in-time supply chains.
That is an example of excessive complexity.
Letwin also draws attention to another issue.
In many cases we have replaced technologies that are somewhat less productive but highly robust with technologies that are more productive and efficient but much less robust.
One recommendation I make in the book is that sometimes it makes sense to pay the cost of keeping the older, more resilient technology available.
For example, you probably don’t want to become completely dependent on mobile telephony because it is vulnerable to both malicious human action and natural events.
There is an argument for retaining old-fashioned landlines and even telephone boxes as a backup because the technology is much more robust.
The other problem connected to excessive complexity is the lack of margin.
In a very real sense—and this may sound strange coming from someone interested in economics—I would argue that the world has become too efficient.
Many systems no longer have sufficient buffers or spare capacity to deal with extreme events.
SpeakFreely: One of your main examples is the NHS.
SD: Yes. The NHS is a classic example.
During COVID it became very clear that the NHS, along with several other major healthcare systems around the world, did not have enough spare capacity.
That applied to hospital beds, but even more importantly to skilled staff.
The system could cope with fluctuations around the normal level of demand.
But when you experienced a tail event—a sufficiently large departure from the normal range—it simply could not handle it.
You can see the same phenomenon in many areas of contemporary economic and social life.
We have created systems that work extremely efficiently as long as circumstances remain reasonably close to the long-run norm, but which cannot handle a major deviation from it.
Imagine I’m the Emperor of China and I hire a brilliant new adviser with an MBA from Harvard.
He looks at all my flood defences and says: “You’re wasting an enormous amount of money. These defences are designed for a hundred-year or two-hundred-year flood. That is an incredibly unlikely event. Cut the expenditure and spend the money on something more productive.”
So I do. Ten years go by and nothing happens. We have ordinary floods and everything seems fine. Then eventually the one-in-a-hundred or one-in-two-hundred event actually happens.
A huge flood devastates North China. And at that point I’m probably not Emperor anymore. That is the problem.
For various reasons connected to investment markets and regulation, we have focused excessively on a very narrow—and statistically innumerate—definition of efficiency.
Paradoxically, you can become too efficient. You become so focused on maximising returns under normal conditions that you eliminate the resilience you need for the occasion when normal conditions disappear.
SpeakFreely: So you can optimise while the game is continuing, but once the game itself disappears, that optimisation no longer matters.
SD: Exactly. With a tail event, sometimes you can recover afterwards. But sometimes the game is simply over. Nassim Nicholas Taleb uses a useful casino analogy. Imagine lots of different people visiting a casino each night. Some will win, some will lose, and some will lose a great deal. But over a sufficiently large number of people and visits, the aggregate results become relatively stable. That’s the law of large numbers. Now imagine instead one person—Joe—going to the casino every night. He wins sometimes. He loses sometimes. But eventually there is a chance he loses everything. Once that happens, there are no more visits to the casino. The game is finished. That is what a catastrophic risk is like. It may be very unlikely, but if the tail event happens and it ends the entire system, there are no more plays.
SpeakFreely: One of the approaches you suggest for protecting against global catastrophic risks draws on the work of Vincent and Elinor Ostrom and their approach to governing commons.
To a lot of people, the commons approach sounds quite complicated.
How could it help us deal with these risks? And is it something government would need to incentivise?
SD: No, I don’t think government needs to incentivise it.
Let me explain the Ostroms’ work, particularly Elinor Ostrom’s.
She was interested in how traditional communities managed common-access resources.
Ever since Garrett Hardin published his famous essay The Tragedy of the Commons in the 1960s, economists have been familiar with the problem.
If everyone has access to a common resource, it can be individually rational for each person to maximise their own use of it even though, collectively, this eventually destroys the resource.
You can see this in things such as global fisheries and fish stocks.
One solution is to privatise the commons—to establish clear property rights.
But you cannot always do that. And in some cases you may not want to, because the common resource is closely connected to wider social systems and institutions.
What Elinor Ostrom showed was that communities can successfully govern common resources without relying either on conventional private-property markets or on top-down government regulation.
They can do it through voluntary, bottom-up cooperation.
Crucially, this is often spontaneous. It doesn’t appear to require government encouragement or intervention. In fact, government interference can sometimes undermine it.
Instead, it tends to depend upon certain social preconditions: high levels of trust, repeated interactions between participants and similar factors.
These arrangements can emerge spontaneously. One of the questions Ostrom was exploring towards the end of her life—and which her students continued to examine—was how these systems could scale.
Many of the case studies she and her colleagues studied were highly localised arrangements, such as Maine lobster fisheries.
The interesting question was whether these local institutions could associate through networks and nested structures until you ended up with systems covering very large geographical areas and populations. But the foundation remains voluntarism, mutual cooperation and bottom-up organisation.
Instead of instructions and knowledge flowing from the top down, you build a structure in which institutions emerge from the bottom up.
SpeakFreely: And how does that relate to global catastrophic risk?
SD: Some global catastrophic risks—climate change is a particularly important example, in my view—may be better addressed through this kind of dispersed approach.
The alternative would be to rely on governments around the world acting collectively. But that may simply be unrealistic because the collective-action problems are so intractable.
You run into prisoner’s dilemmas and other coordination problems. And then there are the downstream costs.
Suppose you assume that the only way to deal with that collective-action problem would be to create a dictatorial world government, and as I said a moment ago, that would actually be worse than the disease. So for a lot of problems involving the way human beings interact with the natural world and with one another socially, I think the Ostrom approach is the best one available to us—the one most likely to pay off.
Is it complicated? Not really. It looks complicated because Ostromian systems have very high levels of redundancy.
To go back to the point I made earlier, one of the problems with many of the systems we have today is that they have very low redundancy. There simply isn’t enough margin.
But Ostromian systems also aren’t based on elaborate, abstract procedural rules.
That kind of rule-following is what often generates excessive complexity: the attempt to organise an activity through increasingly detailed formal procedures that everyone is obliged to follow.
That is the essence of institutional management. And I think we, as a species, are beginning to push up against the limits of that process-driven managerial approach. The approach that Elinor Ostrom and her students identified, and which Vincent Ostrom also explored in his political theory, is different. It isn’t based on highly formalised procedures. It relies instead on fairly general rules that structure spontaneous human interaction. The thing that produces the outcome is the interaction itself, not the process. That allows for much greater variability. So these systems can appear complex, but the underlying operating system, if you like, is actually quite simple. That is what Elinor and Vincent Ostrom called polycentric governance.
SpeakFreely: That’s fascinating. And in the book, you don’t dismiss government altogether. You clearly don’t want a world government because of the tyranny problem, the knowledge problem and the incentive problem. But you don’t argue that government has no role.
SD: No. There are some areas where I think government involvement is unavoidable. I mentioned large-scale solar flares earlier. Another obvious example is the risk of nuclear war, which I think is one of those catastrophic risks that we have tended to forget about and are now seriously underrating.
We should take nuclear war much more seriously than we currently do, given that it would be truly catastrophic. The popular attitude seems to be: we’ve had nuclear weapons since 1945, there hasn’t been a general nuclear war yet, so apparently we’re fine. That is a bit like watching somebody play Russian roulette and saying: “He’s spun the barrel and pulled the trigger ten times without blowing his brains out, so there’s obviously nothing to worry about.” No. It means he has been extraordinarily lucky.
I think that is roughly the position we are in with nuclear war. We also tend to forget how close we came to accidental nuclear war during the Cold War. There were several occasions when we came frighteningly close to a nuclear exchange that could have ended advanced civilisation. As long as nuclear weapons exist, that risk remains. And current political events may be increasing it. Governments around the world may look at the behaviour of the current US administration and conclude that they need their own nuclear deterrent. Iran would be an obvious case. So that is a risk that may be rising.
SpeakFreely: That brings us nicely to your other new book, The Great Realignment.
Does the political realignment you describe make it more difficult for governments—particularly in Britain—to build coalitions and cooperate on these kinds of global risks?
SD: Possibly in the short run.
Let me briefly explain the realignment argument.
My argument in The Great Realignment is that politics used to be organised primarily around economics.
You had one side that favoured less government intervention in the economy and another side that favoured more, with different degrees within those broad camps.
My argument is that this is no longer the main dividing line.
We have moved into a political world in which the major issue sorting politics into left and right is increasingly whether you favour the nation state or some form of cosmopolitan globalism.
Does that make dealing with global catastrophic risks harder? In the short run, perhaps. Political realignments typically create periods of confusion while people sort themselves into new camps. A lot of people discover that those they were previously allied with are now on the opposite side, while people they previously opposed suddenly become allies.
That happens because the new dividing line cuts across the old one. So you get fluctuating party fortunes and unstable governments for a while.
Eventually, though, the new alignment stabilises and you again end up with two reasonably clear sides. You might argue that the nationalist side of this new divide is inherently less willing to engage in the international cooperation required to address global catastrophic risks. For some problems, such as nuclear weapons, where collective government action really is necessary, that could be a serious difficulty.
But for other risks, I think that argument misses the point because government action at the global level was probably never going to be feasible anyway. Abrupt climate change would be one example where I think you need to rely much more heavily on other mechanisms. More broadly, I don’t think the exact ideological content of the political divide at any given moment has that much bearing on whether the state is actually capable of acting to reduce these risks.
The more important question is state capacity. Do contemporary developed states still have the practical ability to act effectively? That question matters not just for global catastrophic risks but across a whole range of policy areas.
One of the most striking features of contemporary politics—in Britain, across much of Western Europe and in large parts of the United States—is how ineffective government has become.
Governments consume huge amounts of resources and spend enormous sums of money, yet often seem incapable of getting things done effectively.
And I think that is another manifestation of the excessive institutional complexity we were talking about earlier.
SpeakFreely: In your realignment argument, where politics is increasingly dividing between nationalist and cosmopolitan outlooks, what are the economic and social reasons people have begun sorting themselves into those camps?
SD: To summarise an entire chapter very quickly, the explanation begins with an economic process that has been unfolding since the collapse of Bretton Woods in 1971.
Increasing numbers of voters have experienced that economic transformation not primarily as an economic phenomenon, but as a cultural and demographic one that they perceive as threatening their identity.
The process behind this is the particular form that globalisation of production and distribution has taken since Bretton Woods. From roughly 1948 into the early 1980s, globalisation was primarily driven by trade. It was largely about removing barriers to international trade. But after the collapse of fixed exchange rates and the ending of exchange controls, you also got globalised capital flows.
So globalisation increasingly became investment-driven rather than simply trade-driven. That produced two major consequences. One concerns agriculture in the Global South. Agriculture moved away from subsistence farming and towards more market-oriented production, with many beneficial effects. But it also made enormous numbers of agricultural workers economically redundant across Latin America, South Asia, Southeast Asia and Africa. Many of those people could no longer make a living in rural areas. Most moved to cities in the Global South, but a significant minority sought to migrate to the Global North. So you get large-scale migration driven in part by this transformation.
There are also pull factors in the Global North, notably ageing populations. The second major consequence is that the world economy is no longer really organised on a national basis.
It increasingly operates as a global system structured around city regions rather than national units.
It makes less and less sense, economically, to talk about “the British economy,” “the Spanish economy” or “the French economy” as completely coherent units.
In Britain, for example, you have the London economy and then, in important respects, the economy of the rest of the UK.
Ireland has something similar with Dublin and the rest of the country. You can make comparable observations in many countries.
At the same time, there has been a major relocation of manufacturing away from Europe, North America and Japan towards the Global South—particularly East Asia, increasingly South Asia, and potentially Africa in the future.
People living in deindustrialised regions have often experienced that shift as a threat to their cultural identity.
They also perceive the nation state, with which they strongly identify, as being weakened and partly replaced by systems of global governance.
Those systems emerged in part because the global economy itself became organised through networks of city regions, coordinated through a relatively small number of major world cities.
This is fundamentally an economic transformation. But large numbers of voters experience it as a process of cultural and social change that threatens the identity of their neighbourhood, locality, nation and, increasingly, their own economic and social position. As a result, they begin prioritising issues such as national sovereignty, immigration and national identity over traditional economic questions. That is what produces this new kind of politics and, in particular, the new forms of right-wing politics we have seen emerging in country after country.
SpeakFreely: You mentioned families and immigration. Another issue we’re facing is declining fertility. Do you regard the fertility crisis as a potential global catastrophic risk?
SD: It could become one if it isn’t addressed. People don’t appreciate just how dramatic the collapse in total fertility rates has been.
There are really two things happening. We have known for a very long time that there is a relationship between economic development, average income and the number of children people have. Once countries become sufficiently wealthy, they tend to go through what demographers call the demographic transition, where fertility falls towards or below the replacement rate of roughly 2.1 children per woman.
We have seen that in country after country. As the world has become more developed, fertility has therefore declined globally. But something else appears to have happened around the 2010s. There was a further sharp downward shift in fertility in one country after another, including quite dramatically in countries that were not yet highly developed.
Parts of the world that had maintained very high fertility rates—the Middle East, large parts of Africa and much of Latin America—also experienced steep falls. There are now very few countries with fertility above replacement, and even many of those are experiencing rapid declines. Because some of these populations are still very young, their total population may continue growing for decades due to demographic momentum.
But underneath that, fertility is falling sharply. People also fail to appreciate that population decline can compound just as population growth does. In my own lifetime, the global population has increased enormously. That upward curve compounds.
But if fertility remains very low for a prolonged period, the downward curve can compound as well. Over the longer term, you could therefore end up with a global population dramatically smaller than today’s and still shrinking rapidly. That could potentially become a global catastrophic risk because it is not obvious that an advanced civilisation of the kind we have now could be sustained with a sufficiently small and rapidly declining population.
SpeakFreely: Very quickly, because we’re almost out of time: have you thought about solutions?
My understanding is that direct government subsidies for having children don’t seem to work particularly well. I’ve written about how housing and land-use regulation can affect fertility, but even reforms there probably wouldn’t restore fertility completely.
SD: No. And the uncomfortable answer is that I honestly don’t know what the solution is. The evidence seems quite strong that measures of the kind Hungary and other governments have tried do not have very large effects. Even if those measures did work somewhat, you might get fertility back towards replacement, but probably not sustainably above it.
Achieving that may require a much more radical transformation in how society is organised. Much of our contemporary economy and social structure is effectively built around people having relatively few children. To reverse that structurally, you might need very substantial changes—for example, perhaps some revival of household production.
And there is another problem. I don’t think we yet properly understand what caused the particularly sharp fertility decline after around 2010. We simply do not have a convincing explanation. There may even be biological factors involved. We really don’t know. So understanding what caused that decline is itself an important part of finding any solution.
SpeakFreely: Steve, thank you so much.
SD: My pleasure.
