Demographics, Growth and Financial Markets

Erik Haller Pedersen, Economics

INTRODUCTION AND CONCLUSIONS

In the coming decades, demographic structures will change, both globally and in Denmark, with steadily more elderly people and fewer people of working age. So far, the debate in Denmark has focused mainly on the consequences of the ageing trend for government finances and the future financing of the existing welfare schemes^[1]. This article takes a slightly different approach by considering the potential consequences of ageing for the financial markets and the economy in general. To examine this topic, it is necessary to look beyond Denmark's borders and view the issue in a global perspective.

The conclusion is that the coming decades will see a radical shift in the global population and the economic power balance between countries and regions, e.g. as a consequence of diverging population growth. The global economic importance of Europe will decline. Ageing populations are a global phenomenon, but the process is more pronounced and further advanced in Europe and Japan than in other parts of the world.

The analysis shows that we are at a demographic turning point. Where demographic trends in recent decades have supported lower interest rates and higher share prices and cash prices for owner-occupied homes, the demographic development in the next 50 years is likely to exert upward pressure on both nominal and real interest rates, and downward pressure on share prices and housing prices. However, it is by no means certain that this will actually result in rising interest rates and falling share prices and housing prices, since the demographic effect will in all cases be dominated by other factors, which may have the opposite effect. It is difficult to find robust historical correlations as the basis for projections.

THE GLOBAL DEMOGRAPHIC DEVELOPMENT UNTIL 2050

According to the UN's medium scenario^[2], the global population will grow from just over 6 billion today to more than 9 billion in 2050^[3], cf. Chart 1. Most of this growth will be seen in Asia and Africa, but the populations of the Americas will also be on the increase. In all regions, the rate of population growth will decline towards 2050, but it will only be negative in Europe, Japan and China, cf. Table 1.

Viewed in isolation, the declining fertility rate means that over a protracted period the average age of the population will increase. The average life expectancy is also increasing globally, which further amplifies this process. Different regions are at different stages of the ageing process, cf. Chart 2. In the fastest-ageing countries, i.e. the high-income countries, ageing typically entails that the large generations from the 1940s reach retirement age^[4], and the population of working age declines. For low- and medium-income countries, on the other hand, large generations reach working age and are replaced by smaller generations as a result of decreasing fertility rates, i.e. initially the percentage of the population that is of working age increases in these countries.

In Europe, the median age in 2050 will be close to 50 years, i.e. half the population will be above this age, cf. Table 1. In southern European countries such as Spain and Italy the median age will be as high as 55 years. The traditional age pyramid will simply be turned upside down.

The ageing of the population will be inevitable in the decades to come, and we must come to terms with it. If the ageing process in Europe were to be resolved purely via immigration from regions with younger populations, this would require immigration on a much larger scale than today.^[5]

THE DEVELOPMENT IN THE GLOBAL ECONOMIC POWER BALANCE

Declining populations in Europe, combined with ageing, will lead to significantly fewer people of working age, which will affect the growth potential of the economy. The EU estimates that its annual potential growth will decline from 2.5 per cent today to only 1.5 per cent in 40 years' time. The reverse situation is seen in regions with growing populations of working age. Consequently, the distribution of global GDP by regions will change drastically over the next few decades. This is not only attributable to demographics, but also to the catching-up effect, whereby GDP per capita in the poorer countries approaches that of the wealthier nations. The exception is Africa, where growth is still slow in many countries.

Chart 3 shows GDP by regions in 2005 and 2050, cf. a projection by Goldman Sachs (2003). Other projections give slightly different figures, but the overall pattern appears to be the same. China and India will in all probability develop into economic superpowers, while the relative size of Europe in particular will decline. This development is natural in a way since it will ensure greater balance between the geographical location of the global population, and thus the workforce, and value creation.

The welfare consequences of the demographic development are, however, linked to GDP per capita rather than to the absolute levels, cf. Box 1. According to the above projection by Goldman Sachs, the "old" industrialised countries will still have the highest GDP per capita in 2050, cf. Chart 4.

DEMOGRAPHICS AND THE FINANCIAL MARKETS

In the economic literature it is typically assumed that people's financial behaviour is determined by a wish to balance consumption over their lifetimes, i.e. consumption and savings decisions are a function of the individual's age, cf. Appendix 1. Young people between the ages of e.g. 20 and 34 are typically borrowers, the middle-aged between 35 and 59 are net savers, while the over 60s are running down their savings (dissavers). Continuing along this line of thought, the dependency ratio can be calculated as borrowers (20-34 years) and those who live off their savings (dissavers) (60+ years) as a ratio of savers (35-59 years), i.e. wealth consumers as a ratio of wealth builders, cf. Barclays (2004). Children and teenagers are omitted since they typically do not make independent savings and investment decisions on any substantial scale. As discussed in Appendix 1, the empirical evidence to support the hypothesis that the over 60s spend their savings is not very strong.

For society as such, the demographic development will thus influence savings and consumption, which in turn affect the level of interest rates and returns on shares. If, as is currently the case, a population is dominated by people aged between 35 and 59, the savings ratio will be relatively high, and thereby also the demand for financial assets. This exerts upward pressure on share and bond prices, i.e. interest rates are low. On the other hand, the level of interest rates influences the savings volume, since a real interest rate can be interpreted as the cost of intertemporal reallocation of consumption.

The opposite applies if the population is dominated by borrowers and spenders of savings. As the baby boomers reach retirement age in the coming decades, they will to some extent seek to realise their assets, which will push share and bond prices down and interest rates up, since the generations that are to take up the assets are smaller. This effect is mitigated if the smaller generations save relatively more than previous generations, which is presumably the case, e.g. in view of the international trend for increased funding of pension savings.

On the face of it, the correlation between the dependency ratio and the nominal interest rate is surprisingly strong, cf. Charts 5 and 6. The high level of interest rates in the 1970s coincided with a relatively small population aged 35-59. As the baby boomers have reached the age where they save up, the level of interest rates has declined, and in recent years it has reached a low, while the group of savers is back at the level seen in the 1st half of the 20th century. Looking ahead, the proportion of the population that are savers will decline significantly, and the dependency ratio will therefore increase. This correlation is seen in virtually all industrialised countries and indicates downward pressure on bond prices and thereby a trend for rising interest-rate levels in the coming decades.

Demographic factors are not the classical explanation for the period of high interest rates in the 1970s, nor for the current low level of interest rates. Indeed, interest-rate levels are determined by many other, potentially more important, factors such as monetary and fiscal policy, and furthermore the impact of demographics on interest rates is indirect via savings and investment decisions, which in turn are influenced by many other factors.

As Charts 5 and 6 show, interest-rate levels in recent decades have been in line with demographic developments. However, in theory a shift in the savings-investment balance should primarily be reflected in real, rather than nominal, interest rates, and in this respect the correlation with demographic variables is less apparent, cf. Chart 7. In the 1st half of the 20th century, prices were strongly influenced by World Wars I and II, as well as the deflation in the 1930s. This resulted in highly volatile real interest rates^[6].

As regards the stock market, the dependency ratio and a real share price index should fluctuate in opposite directions. A faint relationship of this kind is seen in Danish figures for the 2nd half of the 20th century, but the real returns on share investments are highly volatile, cf. Chart 8.

Potaba (2001) analyses relations between demographics and financial returns using US data and concludes that it is difficult to find a robust correlation between demographics and real returns, including on shares. The demographic variable is overshadowed by other factors, even in the long term. Consequently, interest rates and share prices cannot be projected solely on the basis of demographic variables. It does not seem realistic that the level of interest rates towards the middle of this century will reach the level from the 1970s, even though the dependency ratio does, but demographics are likely to have a qualitative impact in supporting rising interest rates.

Even if the correlation between demographics and real returns is weak, the question remains of whether there is a link between demographics and inflation. The argument would be that the demographic development can be more or less favourable in relation to inflation. For instance, when inflation became endemic in the 1970s, one of the reasons was, according to this argument, a small population of working age. A demand shock such as the oil price hikes thus triggered a price and wage spiral^[7]. Again, this is not the traditional explanation, which typically attributes high inflation to accommodating monetary policy combined with unsuccessful fiscal policy. In Denmark's case, foreign-exchange policy was also a factor, since repeated devaluation stimulated the wage/price spiral.

That demographic factors play a role in price development is evident today, where a large labour reserve in e.g. Asia exerts downward pressure on global consumer prices. Looking forward, we might take a more politological approach. As the electorate ages, economic preferences for low inflation will become stronger. Based on this argument, the demographic development over the next 50 years will point to low inflation. All things considered, it is, however, problematic to view demographic fluctuations as the decisive factor behind variations in inflation.

A fundamental weakness of correlating demographic variables with financial variables, inflation, etc. is that the demographic variable is very sluggish. Even with figures dating back to the early 20th century, the degrees of freedom are in fact very limited. This makes it difficult to perform convincing econometrics on the data for use as a basis for projections, cf. Potaba (2001).

DEMOGRAPHICS AND THE HOUSING MARKET

Homes are traded internationally to a far lesser extent than financial assets. Consequently, purely national factors are more likely to influence price formation. Chart 9 shows demographically determined demand for owner-occupied homes, determined on the basis of a breakdown of Danish cash prices for owner-occupied homes in 2002 by age groups, cf. Appendix 2. The demographically determined housing demand shows the isolated impact of demographics on housing demand, all other things being equal. On the basis of a demographic projection, the resulting housing demand in constant prices can be projected up to 2050 and compared with the historical development in real cash prices^[8].

As the Chart shows, the demographic development over the last 50 years has led to higher demand for owner-occupied homes and, viewed in isolation, this has supported the increase in cash prices. This will change in the coming decades as the demographically determined demand flattens and then falls a little. On the other hand, there does not seem to be any basis for fearing a demographically determined collapse of cash prices in the coming decades, but unlike at present the demographic development will dampen the increases.

The above should not be viewed as an estimate of the future development in real cash prices, since they are influenced by many other factors besides demographics. Firstly, real incomes will presumably continue to rise, which in itself will increase the housing demand – in terms of more homes, but also in terms of the quality of existing homes. Secondly, interest rates and economic development are also important, especially in the short and medium term. For instance, the real cash price in Chart 9 is strongly influenced by the economic stop-go policies of the 1980s, as can be seen. Housing demand is also affected by a steady fall in the size of the average household, as the number of singles increases. This boosts the demand for housing from a particular population group. Finally, the supply side is naturally also of paramount importance in the long term.

What remains, however, is that a significant factor underlying housing demand, viz. population growth, will decline in the coming decades, while at the same time the age structure will change. Viewed in isolation, this will dampen the growth in cash prices. On the other hand, there are no indications of a demographically determined collapse of the housing market. Large regional differences are, however, to be expected.

APPENDIX 1

As expected, US data shows a clear negative correlation between the dependency ratio and the households' wealth, cf. Chart 10. When the 34-59-year age group is large, as is currently the case, the savings ratio – and thus the wealth ratio – is high, and vice versa. It is significant to the result that wealth gains are included. Applying the savings ratio from national accounts, which does not include capital gains, the correlation is positive, not negative. In this article, the USA is used as a proxy for the rest of the world.

In the life-cycle hypothesis, the theory is that people save while they are active in the labour market and then spend their savings when they grow older. The question is, however, whether the theory of older people as wealth consumers holds water. Potaba (2001) shows that it is doubtful for US data, and Danish data do not seem to support the theory either, cf. Table 3. It is true that the net wealth of the over 85s is considerably lower than for other retired people, but the main explanation is that relatively few in this group own their homes. However, it might be significant that pension wealth is omitted. If it is included, the conclusion may be less clear.

APPENDIX 2

On the basis of cash values broken down by age groups, the total demand of each age group is calculated, defined as the cash value multiplied by the number of households in the age group. This figure is multiplied by an index of the development in the population from 1955 to 2050, with index 2002 = 1. This gives the demographically determined demand for owner-occupied housing within the age interval in question. The aggregate housing demand is found by adding the age groups. This calculation implicitly assumes that the relative housing demand across age groups is constant over time. For a further discussion of the method, see Mankiw and Weil (1989).

REFERENCES

Abildgren, Kim (2004), Nominal and real effective krone rate indices for Denmark 1875-2002, Danmarks Nationalbank, Working Paper no. 13.

Abildgren, Kim, (2005), A historical perspective on interest rates in Denmark 1875-2003, Danmarks Nationalbank, Working Paper no. 24.

Barclays (2004), Global Speculations, Dismal Demographics, Barclays Capital Research. www.barcap.com.

Mankiw, N. Gregory and Davis N. Weil (1989), The baby boom, the baby bust and the housing market, Regional Science and Urban Economics 19.

Goldman Sachs (2003), Dreaming With BRICs: The Path to 2050, Global Economics Paper No: 99.

Pedersen, Erik Haller (2001), Development in and Measurement of the Real Interest Rate, Danmarks Nationalbank, Monetary Review, 3rd Quarter.