How ageing populations will affect interest rates

Our history is our database. When seeking to peer dimly into the future, our normal response is to examine what happened in (similar) past episodes and then to extrapolate those outcomes into the future. This assumption, that the future will mimic the past, is hard-wired into almost all our forecasting exercises, from the most simple to the econometrically and technically most complex.

Yet this assumption, that the future will be like the past, is perhaps more questionable now than for decades, at least in the economic sphere. All around the world we stand on the cusp of a dramatic shift in the structure of our populations, the ageing of our people. Only Japan has yet decisively entered into this difficult new world, and its experience has been affected by some special factors; it coincided with a financial crisis and occurred while its neighbours in Asia were still benefitting from a population ‘sweet spot’, with their ratio of workers to dependents (otherwise known as the ‘support ratio’) rising. So its experience is not necessarily a reliable guide to the future either.

Future demographic changes

In the standard demographic transition process, outlined by Ronald Lee in his paper at this year’s Jackson Hole Conference (Lee 2014), with improving living and medical conditions,

“mortality begins to decline from its initially high level, followed typically some decades later by the beginning of fertility decline from its initially high level. During this early phase the population growth rate first rises and then declines, and the share of the population in the working ages first declines and then rises. Inevitably low fertility and mortality lead eventually to population aging as a final outcome. But population aging is substantially delayed, starting decades after fertility begins to fall. Even in Japan, which is currently farthest along in the aging process, aging is still at an early stage. The old age dependency ratio (ratio of population 65 and over to population 20 to 64, or OADR) will be twice as great in 2050 as in 2010, rising from .39 to .78 according to United Nation projections.”

Most of the world – excluding Africa and, perhaps, parts of Latin America, but including Asia, Australia, Europe, and North America – is now at the point where the support ratio, defined as the ratio of producers to effective consumers, shifts sharply from being beneficial to being adverse, as shown in Tables 1 and 2, taken from Lee’s same talk. Not only is the support ratio falling, but also the absolute number of those of working age (taken to be 20–65) will decline in many countries, and be much lower in the next 35 years (2015–2050) than in the past 35 years (1980–2015).

Table 1. Support ratios have peaked and will fall in emerging markets

1950 1970 1990 2010 2030 2050 2100 China 1.13 0.99 1.14 1.34 1.24 1.11 1.05 India 0.76 0.72 0.75 0.82 0.88 0.87 0.78 Brazil 0.54 0.52 0.58 0.67 0.67 0.60 0.50 Indonesia 0.64 0.63 0.66 0.77 0.81 0.82 0.77 Thailand 0.64 0.63 0.75 0.90 0.83 0.71 0.65 Mexico 0.52 0.47 0.52 0.61 0.64 0.62 0.54 Chile 0.66 0.60 0.70 0.75 0.73 0.67 0.59 South Africa 0.66 0.61 0.63 0.71 0.76 0.78 0.73 Turkey 0.70 0.68 0.72 0.81 0.85 0.82 0.74

Source: National Transfer Accounts

Table 2. …and are nose-diving in developed markets

1950 1970 1990 2010 2030 2050 2100 US 0.79 0.71 0.78 0.80 0.72 0.70 0.65 Japan 0.66 0.78 0.83 0.76 0.67 0.58 0.57 Germany 0.84 0.76 0.84 0.81 0.68 0.62 0.59 UK 0.82 0.74 0.77 0.77 0.70 0.67 0.62 France 0.85 0.77 0.83 0.83 0.74 0.72 0.67 Italy 0.70 0.71 0.74 0.77 0.67 0.60 0.59 Sweden 0.98 0.91 0.91 0.91 0.83 0.80 0.73 Canada 0.71 0.67 0.78 0.82 0.73 0.69 0.65 Australia 0.80 0.74 0.80 0.80 0.73 0.70 0.64

Source: National Transfer Accounts

In so far as we can assess the economic implications of the generally beneficial demographic changes of the last 35 years, this would give us a reasonable starting point for examining how the reversal of these changes will affect our economies over the next 35 years. In particular the past beneficial (future adverse) shift in the support ratio will have made (will make) output and consumption per head grow faster (slower) even with no changes whatsoever in output growth per worker.

Similarly, the slower growth in the number of workers must slow down the absolute growth rate (comparing 2050–2015 with 2015–1980), if there is no change at all in the growth of output (productivity) per worker at all. Some illustrative figures on working population are provided in Table 3.

Table 3. Growth in the workforce (%)

1990-2000 2000-2013 1990-2013 2013-2050 Japan 3.1 -2.1 1.0 -33.7 US 15.2 5.1 21.2 18.0 UK 2.3 8.8 11.3 1.8 Germany 1.7 6.3 8.1 -34.3 China 11.3 6.8 18.9 -10.5

Source: Haver Analytics, Morgan Stanley Research

Of course, the population projections (e.g. future fertility and mortality rates) can change. Whereas (nuclear) war and pestilence could raise mortality rates, so far they have consistently surprised on the downside, and the human genome project could extend that trend. As women get better education and control of their own destinies, fertility rates decline. In each individual country, population trends also depend on migration patterns, but we shall not discuss that here, simply taking the UN figures as the basis of our exercise.

Some economic implications of a declining workforce

So, what are the other implications of the reversal of the recent beneficial population trends? Assessment of this is made much more difficult by a couple of facts. First, the increase in support ratios and growing working populations have gone hand-in-hand with increased globalisation, so that entrepreneurial skills, technical skills, and capital sourced in countries in the developed world could be applied in production and labour employment elsewhere, notably in the emerging countries. Second, the Asian countries where both the population and globalisation effects were strongest (e.g. China, South Korea, Malaysia, Thailand, Vietnam, and Singapore) were generally well administered, with well-educated and skilled labour forces. Distinguishing a general population effect from globalisation and specifically Asian effects is not easy.

Of what population effects are we reasonably confident? First, the higher (lower) the support ratio, the higher (lower) will be the average household savings ratio. Apart from low consumption in the first few years of life, consumption is reasonably constant over the life cycle, while income is earned, and output produced, in the working life between 20 and 65. As Figure 1 suggests, the old dis-save, while workers save. The more old people there are, the less saving there will be.

In the US, the old actually now consume more than the rest, largely due to medical expenses. A particular problem of ageing is that many conditions make them helpless (e.g. dementia), so that besides drugs, hospitalisation, etc., they need continuous care from others – often those of working age.

The financing of old-age consumption varies greatly from country to country. One of the benefits of the past improvement in the support ratio was taken in a slight lowering of the retirement age. Table 4 shows the retirement age (men and women) in selected countries in 1990 and 2014. What will it be in 2050? If we were to assume no change in transfers (from workers to old), and equal consumption of old and workers, it would arithmetically have to be about 70.

Figure 1. The old consume, workers save

Source: National Transfer Accounts

Another way of handling the prospect of an older population would be to have more of them participating in the labour force, a development that improved standards of living and medicine should facilitate. The distinction between participation (in the labour force) rates in Asia, where public sector pensions are generally low, and in developed countries in the West, is marked (see Figure 2).

Figure 2. Higher pensions are associated with lower employment of the old

Source: OECD

Countries whose support ratios are becoming better than average tend to have higher household savings ratios, and with ex ante savings greater than ex ante investment tend to run current account surpluses. Examples are Japan, up to about 1980, and China and South Korea now. The purchase of assets abroad, the accompanying capital outflow, can be reversed as the support ratios decline. This has already happened in Japan, and is happening in China and South Korea, and may well also happen in Germany.

Table 4. Average effective retirement age

1970 1990 2012 US 68.5 64.7 65.0 Japan 72.3 70.4 69.1 UK 67.7 62.8 63.7 France 67.6 60.0 59.7 Italy 65.0 61.8 61.1 Spain 69.4 62.9 62.3 China – 66.9 66.8 Mexico – 77.8 72.3

Source: OECD

The international monetary system has struggled to cope with the persistent surpluses of China, Germany, and the oil-producing countries, matched with deficits in the US and the UK. With the latter having relatively more favourable demographic dynamics than China or Germany, the previous pattern of surpluses and deficits is likely to change. If, and it may be a big if, the subcontinent (Bangladesh, India, and Pakistan), Indonesia, and Africa can improve their governance and the educational skills of their people, they are best placed to reap the benefits of demographic dynamics – growing faster, with higher savings ratios, and current-account surpluses. North-East Asia will move into current-account deficit, as will continental Europe. The Anglo-Saxon world – the UK, North America, and Australia/New Zealand – will be more balanced. Africa, Latin America, and the subcontinent could move into sizeable surplus.

What is much more uncertain is what will happen to inflation, business investment, interest rates, both real and nominal, and to output per worker.

There may be cross-currents affecting the future course of inflation. On the one hand a slower growth of workers, and of population, will take some of the pressure off land and natural resources. Moreover, the urge to limit global warming should maintain sufficient technical progress in renewables, probably mainly solar, to lower the price of hydrocarbon fuel sources. On the other hand, the decline in the availability of labour – and the need to impose ever higher taxes on workers to shift resources to the growing army of the old – is likely to reverse the trend reduction in the power and share of labour in national income, which was the key theme of our previous column (Erfurth and Goodhart 2014). Overall we see no good reason why central banks should fail to continue to achieve long-term inflation targets, set at about 2%, unless we have underestimated the persistence of deflationary pressures over the next decade or so.

Let us turn next to investment. A slower growth of population might be assumed to lead directly to lower residential investment over the next 35 years. We rather doubt that. The income elasticity of housing is high; more space and second holiday homes. As incomes rise, the young and the old would rather live in separate dwellings than with their parents/children. What will happen to business investment? By and large, business investment has been lower than expected/desired in most countries over the last couple of decades. Could this have been in part caused by the relative cheapness and availability of labour, especially the option of outsourcing abroad? If labour cheapens, one would expect the K/L ratio to decline. A theme of this column is that such cheapening has, perhaps, reached its strongest point, and may now go into reverse. If so, one might expect the K/L ratio in business to start to rise again.

Wicksell, and others, established that the natural (or normal) real rate of interest is a product of thrift and productivity. Ageing populations almost by definition will be less thrifty; the household savings ratio will fall. What we have argued above is that the ex-ante desire to invest may fall somewhat, but almost certainly less than the ex-ante desire to save. The obvious, indeed almost inevitable, conclusion is that real rates of interest will reverse from their present decline, and go back up. The current negative real rate of interest is not the new normal; it is an extreme artefact of a series of trends, several of which are coming to an end. Where might real interest rates reach? By 2025 they should have returned to the historical equilibrium value of around 2.5–3%, with nominal rates therefore at 4.5–5%, perhaps somewhat higher by 2050.

Future overall growth rates?

The key question is what will be the rate of growth of output per worker over the next 36 years, as compared with the 2.3% per annum that occurred from 1990 to 2012. The honest answer is that we do not know, of course. But our base case is that it will remain the same, with one exception. The exception is that the explosive growth of China in recent decades will revert to the norm.

We are aware of the various arguments for future slower growth, or even for stagnation, as set out in Gordon (2012, 2014). And not only is the argument that there could be a slowdown in technological innovation and factor productivity directly challenged by some (e.g. Fernald and Jones 2014), but there are also two other counter-arguments. First, the combination of a demographic sweet spot, plus the entry of China and Eastern Europe into the world economy, made labour unusually and increasingly cheap in recent decades (see our previous note here). The expected reverse tightening of the world’s labour markets will enhance the future incentives of management to hold down unit labour costs by raising labour productivity faster than in the past. Second, whatever the rate of innovation at the leading edge of science, IT and the internet will spread existing technology much more rapidly around the world.

Where we are more confident is in predicting that, given the rate of growth of output per worker, the rate of growth of output per head will slow, as the ratio of workers to population slows, and the aggregate rate of growth will also slow, though by rather less, as population slows. All this is shown diagrammatically in Figure 3.

Figure 3. Output per head will trail output per worker

Source: Haver Analytics, Morgan Stanley Research Forecasts

In 1990, the average consumption per head in the world was $4,400; in 2012 it was $5,800. With the same consumption/output ratio as in 2012, and our assumed growth rate of output per worker, population, etc., it would average $12,800 in 2050, compared with $43,850 in the US, $9,450 in Brazil, and $1,050 in India today. If we take as our ‘bull’ and ‘bear’ cases +/- 0.5% p.a. faster/slower growth, then the average level of consumption in 2050 would be $15,400 and $10,600, respectively.

Partly because of our argument that the dissemination of technology will rise in future relative to the leading edge, we tend to believe that there will be some regression to the mean for future growth in output per worker, except for some countries such as Indonesia, which should, given good governance, accelerate, as Table 5 suggests.

Table 5. Output per worker, % avg. annual increase

1990-2000 2000-2013 1990-2013 2013-2050 OECD Author Japan 0.8 1.1 0.9 1.6 1.3 US 2.0 1.4 1.6 1.4 1.6 UK 2.8 1.0 1.7 1.5 1.5 France 1.3 0.8 1.0 1.8 1.0 Italy 1.5 0.2 0.7 0.4 0.7 Canada 1.6 0.7 1.0 1.7 1.3 Australia 2.1 1.0 1.5 2.1 1.5 China 9.3 9.3 9.4 6.6 5.2 India 5.1 6.5 5.9 4.8 6.5 Brazil – 1.7 – 3.0 3.2 Russia – 4.4 – 3.8 2.5 South Africa – 3.0 – 1.9 2.5 Indonesia 3.0 3.8 3.4 4.0 4.5 Turkey 2.8 4.2 3.4 2.5 2.8

Source: OECD, Haver Analytics, Morgan Stanley Research

Conclusions

All around the world we stand on the cusp of a dramatic shift in the structure of our populations, the ageing of our people. Two key changes will deeply influence our economic landscape in the future. First, most of the world is now at the point where the support ratio, defined as the ratio of producers to effective consumers shifts sharply from being beneficial to being adverse. Second, we will observe a slower rate of growth in the number of workers globally.

Both of these changes will have profound and, in our view, negative effects on economic growth globally. A worsening of the support ratio will lead to lower savings, as the old consume more and will make up a higher share of the overall population. Similarly, the slower growth in the number of workers must slow down the absolute growth rate. Beyond these factors, the current global patterns of surpluses are likely to change.

The almost inevitable conclusion is that real rates of interest will reverse from their present decline, and go back up. The current negative real rate of interest is not the new normal; it is an extreme artefact of a series of trends, several of which are coming to an end. Where might real interest rates reach? By 2025 they should have returned to the historical equilibrium value of around 2½–3%, with nominal rates therefore at 4½–5%, perhaps somewhat higher by 2050.

Authors’ note: Charles Goodhart is a Senior Economic Consultant to Morgan Stanley. Philipp Erfurth is a European Economist at Morgan Stanley. This article is not an offer to buy or sell any security/instruments or to participate in a trading strategy. For important current disclosures that pertain to Morgan Stanley, please refer to the disclosures regarding the issuer(s) that are the subject of this article on Morgan Stanley’s disclosure website.

References

Erfurth, P and C A E Goodhart (2014), “Monetary policy and long-term trends”, VoxEU.org, 3 November.

Fernald, J and C Jones (2014), “The Future of U.S. Economic Growth”, NBER Working Paper 19830.

Gordon, R (2012): “Is U.S. Economic Growth Over? Faltering Innovation Confronts the Six Headwinds”, NBER Working Paper 18315.

Gordon, R (2014), “The Demise of U.S. Economic Growth: Restatement, Rebuttal, and Reflections”, NBER Working Paper 19895.

Lee, R (2014), “How Population Aging Affects the Macroeconomy”, Presentation given at the 2014 Jackson Hole Conference, 22 August.

Published in collaboration with VoxEU

Author:Charles A.E. Goodhart is an Emeritus Professor in the Financial Markets Group, London School of Economics. Philipp Erfurth works for the Global Economics Team at Morgan Stanley.

Image: An elderly man monitors share prices on a television screen during a trading session inside the Karachi Stock Exchange December 20, 2010. REUTERS/Akhtar Soomro