Changing Demographics - December 2023 Update
If your customer base is aging with you, then eventually you are going to become obsolete or irrelevant. You need to be constantly figuring out who are your new customers and what [you need to do] to stay forever young.
I wrote the first paper on Changing Demographics in January 2021 (available on request). Has anything changed since, apart from all of us getting about three years older? A fair bit, actually, which is why this update comes quite handy.
It is probably fair to say that a trend which was already on my radar screen in 2021, is now blatantly obvious, and I am talking about the shrinking workforce and the impact of that. It is declining at a very predictable rate and with enormous consequences to follow. My father-in-law has just spent the better part of a week in hospital in Denmark, and the family had to bring him food every day, as the hospital was so understaffed that they couldn’t look after him properly (served yoghurt to him three times a day), and that is probably the most positive feedback I can share with you from a truly disgraceful stay.
The topic of Changing Demographics is enormous, but I shall try and constrain myself to the most important issues. They are, in my opinion:
- The fertility rate – why do women give birth to fewer and fewer children?
- Workforce participation – why do more and more people leave the workforce before reaching retirement age? This is, strictly speaking, not (explicitly) caused by changing demographics, but it has a rather dramatic impact on a growing problem caused by ageing.
- The rising impact of technology – given the shrinking workforce, what is the likely impact of AI and advanced robotics on economic growth going forward?
Those three questions will all be addressed in the following.
The demographic outlook as it stands
People aged 65 and over now comprise the world's fastest growing age group. That age group has begun to outnumber children under the age of five, and, by 2050, they will also outnumber those aged 15-24 (source: UN). You get a pretty good picture of the ageing of the world’s population by looking at a chart depicting the change of the median age over time. By looking at those numbers by country, you can also see which regions are ageing the fastest. That is precisely what you can see in Exhibits 1a-b.
Pretty much the entire world (ex. Africa) has aged over the last 50 years with Europe being hit the hardest, closely followed by North America. What is less obvious when looking at Exhibits 1a-b is that some Asian countries are even worse off than the worst affected countries in Europe. I suggest you fast-forward to Exhibit 2a. As you can see, Asia is well represented in the group of ten countries, suffering from the lowest fertility rate worldwide (implying that the median age is rising).
The workforce will continue to shrink, as we move further into the 21st century. According to the UN, the global population will grow from modestly above 8 billion at present to 10.4 billion in 2100. The UN forecast is based on an assumption that the global fertility rate will remain unchanged vis-à-vis the current level of 2.3. However, I can virtually guarantee you that the fertility rate will continue to decline (more on that later). Therefore, I am of the opinion that overpopulation will soon become a thing of the past; that there will never be more than 10 billion people on this planet. I am not even certain we will ever pass 9 billion.
Why the fertility rate will continue to decline
With a few exceptions, the most noticeable one being Africa, the fertility rate is already well below the replacement rate (2.1) in most parts of the world (Exhibits 2a-b). According to the UN, there are about 1.25 billion people living in Africa today, and that number is expected to grow to 4.2 billion by 2100. In other words, if it weren’t for Africa, the world would begin to shrink, population-wise, within a few years.
This year, the fertility rate in Africa is 4.155, down from 4.212 last year or -1.35% year-on-year (source: macrotrends.net). In other words, the decline in Africa has already begun, although it is still early days. When I argue that the UN forecast for 2100 is way too high, it is down to the fact that the circumstantial evidence is convincing; that higher living standards always lead to lower fertility rates. Take a look at Exhibit 3. As you can see, not only is the global fertility rate only marginally above the replacement rate. It was dramatically higher not that many years ago.
There are indeed a number of reasons why I am convinced the UN is getting it wrong but, at the end of the day, it can be explained by rising living standards which, to a significant degree, are the consequence of urbanisation. In rural parts, children are a financial asset, as they provide an extra pair of hands on the farm. As the family migrates to an urban area, the children become a financial liability. The more children, the more m2 shall be required, the more food must be cooked, etc., etc.
With urbanisation follows easier access to contraception, better healthcare, higher educational standards and more women in the workforce, all of which will have the effect of lowering the fertility rate. In low-income societies in poor countries, it is quite normal to have many children, partially because you know some of them will not survive childhood, and partially because you will need them to look after you as you age. Much of this problem goes away with the higher living standards that an urban lifestyle typically brings. Therefore, I am certain that the 4.2 billion forecast for Africa for 2100 is not going to materialise. Neither do I believe we will ever be more than 10 billion worldwide.
Some thoughts on workforce participation
Globally, both male and female workforce participation rates have declined since 1990 – about 5 percentage points amongst men and about 3 percentage points amongst women (source: ILOSTAT). That said, high-income countries have done better than that. They registered an increase in female participation of about 4 percentage points between 1990 and 2018. Contrary to that, EM countries registered a decline in female participation of about 5 percentage points over the same period.
Female participation in high-income countries has now peaked. It happened in 2017 and is expected to decline by approx. 2 percentage points between now and 2030; however, it is not expected to disturb the ongoing convergence between male and female workforce participation, as the average participation rate amongst men is declining even faster. In Exhibit 5 below, you can see how various regions in the world differ as far as workforce participation is concerned.
This is happening at a very inopportune time. The workforce is already shrinking as a result of the ageing of society at large. If, on top of that, the workforce participation rate is also declining, an already challenging situation becomes even harder to deal with. Earlier, I mentioned the situation in Danish hospitals, but Denmark is far from the only country suffering from a lack of (qualified) labour. Unless well-off OECD countries open their doors to foreign labour, the situation can only get worse. This will not only impact public service levels but also – and that is important – inflation, as wage demands will accelerate in line with the growing shortage of workers.
To wrap up this part, I should probably also mention COVID-19. The pandemic had many effects on society, one of which was a change in lifestyles. The numbers behind Exhibit 3 and 4 above are all pre-COVID. As the pandemic unfolded, many people got used to a different lifestyle, and quite a few continue to pursue that lifestyle. Some work mostly from home now, whereas others have left the workforce altogether. Only time can tell if these people will return to the workforce when their savings, much of which were accumulated during the pandemic, have been fully spent.
The impact of technology
AI will probably affect the labour market more dramatically than other tech innovations have done in the past. Up to this point, few have lost their job when a new technology has been introduced, provided the workers in question have been prepared to undergo re-training. AI is different, as it affects workers at all levels – from those working on the factory floor to university-educated managers.
Having said that, I seem to belong to a minority with that view. In a survey conducted earlier this year by CEPR, almost two-thirds of those surveyed were of the opinion that the rollout of AI will have no impact on the unemployment rate (Exhibit 6). And, when CEPR weighted the responses by the experts’ self-assessed confidence (the chart is not shown), the percentage of those who believe AI will have no impact on unemployment rose to 70%.
CEPR also asked what impact AI will have on GDP growth, and the consensus was that the impact will be substantial and positive and, in this instance, I would agree. On a weighted basis, almost 80% were of the opinion that AI will result in accelerating GDP growth to the tune of 4-6% year-on-year (see here for the survey).
Going back to Economics 101, we know that total output in an economy equals the number of hours worked multiplied by the output per hour (aka productivity). If we assume that the size of the workforce is a good proxy for the number of hours worked (and it is), we know that GDP equals Workforce times Productivity. Using simple math, we therefore also know that:
It is a given that the workforce will shrink for many years to come, i.e. ΔWorkforce will soon turn negative if it hasn’t already happened. Therefore, we also know that the economy will only grow if advanced robotics, AI and other new technologies can be deployed in a way that would make productivity grow faster than the workforce is declining.
As you can see in the CEPR survey referred to earlier, the majority of those surveyed seem to be of the opinion that this is precisely what is going to happen; that the rise in productivity will be sufficient to secure continued economic growth. I don’t necessarily disagree, but I do think it is a bit premature to draw such a bold conclusion.
The jury will probably be out for quite a while on this issue. AI brings many unresolved, legal issues, most of which must be addressed before any firm conclusions can be reached, and that will take time. Skirmishes with unions, who will seek to protect their members from unemployment, could also limit the rollout – at least in the early years. And such conflicts could grant a significant competitive advantage to countries where unions are less powerful.
A few other observations
So far, I have deliberately stayed clear of making too many observations on regional differences, but it ought to be said that the fertility rate varies a fair bit from country to country, and that senior citizens in some countries are far more likely to delay their retirement than in other countries. In that context, the US and the EU are opposite poles. Whereas more and more Americans work well into their 70s, many Europeans continue to demand shorter work weeks, although they are desperately needed.
I can therefore comfortably conclude that, in the years to come, ageing will most likely become a much bigger issue in Europe than it will in the US. Furthermore, if my analysis is correct, GDP will continue to grow faster in the US than it will in Europe, where public healthcare will come under serious pressure.
The two oldest societies in Asia, South Korea and Japan, have a demographic profile in line with the worst affected European countries, but the attitude is (often) different in that part of the world. Whereas in most of Europe, the prevailing attitude is to get as much out of the government as possible, the same is not always the case in Asia.
Furthermore, I expect Japanese and South Korean companies to benefit more from all these new technologies than European and North American companies, and there are two reasons for that:
(i) Asian unions are likely to be less uncooperative than European unions.
(ii) South Korea already has a massive lead in the race to implement advanced robotics, and Japan is number two worldwide (Exhibit 7). This will grant the two Asian countries a massive, competitive advantage in the years to come.
There are many investment opportunities arising from Changing Demographics, the most obvious one being healthcare. Data from the NHS in the UK suggests that an 80-year old man is 6-7 more expensive to the public healthcare system than a 30-year old man. It is therefore a given that healthcare costs will spiral as the population ages.
Another opportunity to take advantage of in the context of Changing Demographics is the change in habits as consumers age. Generally speaking, older people spend less on goods and more on services (e.g. cruise holidays), but they also spend their money differently when buying goods. For example, they spend less on the home and more on goods that can be enjoyed instantly. In that context, it is worth noticing that the biggest buyers of Harley Davidson motorbikes are men in their mid to late 50s.
A third opportunity not to be ignored has to do with the dynamics surrounding South Korea and Japan, which I mentioned earlier. Because South Korea and Japan are two very old societies which shall depend on modern technology like industrial robots to continue to produce what they produce mostly with human labour today, they are ahead of the game in terms of installing those robots. And that has handed those two countries a meaningful competitive advantage. Therefore, one could argue that South Korea and Japan are about to turn what at first glance looks like a disadvantage (i.e. ageing) into a significant advantage – at least as far as corporate profitability is concerned.
The last demography-related dynamic I will bring to your attention today has to do with the tendency for investors to lower their allocation to equities whilst increasing it to bonds as they age. This dynamic was temporarily ‘suspended’ during the years of 0% interest rates, but ageing investors’ appetite for bonds has returned. All things being equal, this is bad for equities – particularly for US equities as Americans are far more active equity investors compared with retail investors in other parts of the world.
It is admittedly hard to quantify the estimated impact of this dynamic but, as long-term readers of my work will be aware, I am of the opinion that US wealth-to-GDP will have to come down in the years to come. Now, if US equities struggle to maintain current valuations as a result of millions of retail investors reconstructing their portfolios in favour of bonds, this dynamic could indeed be one of the factors that drive wealth-to-GDP down in the US.
Niels C. Jensen
15 December 2023