Population decline postponed in Europe

By Juha Alho ( 1 ), Maarten Alders ( 2 ), Harri Cruijsen ( 3 ), Nico Keilman ( 4 ), Timo Nikander ( 5 ), Dinh Quang Pham ( 4 )

Demographers and statisticians from three European countries have analysed the demographic development of 18 European countries, which we shall denote as EEA+ countries. The group consists of the 15 members of the European Union pre-2004, plus Norway, Iceland, and Switzerland. Except for Switzerland, these countries made up the so-called European Economic Area, hence EEA+. (Liechtenstein has not been included.) We analysed historical developments in fertility, mortality, and migration, and we separated the time trend in these variables from random deviations from the trend. By giving estimates both for the time trends and the random deviations in the future, we were able to calculate forecast results in terms of probability distributions, thus quantifying forecast uncertainty. The forecast horizon was 2050. Many methodological details can be found at the web site http://www.stat.fi/tup/euupe/ . The web site also contains forecast results for each of the 18 countries, including details of age and sex for ten-year intervals to 2050.

Based on our forecast, we expect a modest annual growth rate of 0.2 per cent for the population in the EEA+ countries in the years 2003-2050. This contrasts with recent predictions by the United Nations and Eurostat. In its 2004 Revision of World Population Prospects, the UN predicts that the population in the EEA+ countries will decrease in the years 2030-2050 from 407 million to 400 million, after an initial increase from the current level of 392 million (United Nations 2005). In contrast, the new forecast anticipates 427 million inhabitants by 2050. Eurostat predicts that the 15 member countries of the former EU will have a population of 384 million in 2050, 7 per cent less than the new forecast. ( Eurostat’s forecast ). The new forecast assumes higher life expectancy and higher immigration levels than both the UN and Eurostat. However, population growth is modest because assumed future fertility levels are low, ranging from an expected 1.4 to 1.8 children per woman in the 18 countries. This will lead to an eventual decline, but later than has been thought until now.

Strongest growth in the UK, but German population remains largest

Although the population in the EEA+ countries as a whole is expected to grow, some countries may show a decline. Germany, whose 82 million inhabitants account for 21 per cent of the population in the area, is expected to remain the largest in 2050 but its population is expected to fall to 79 million (19 per cent) in 2050. However, France will increase from 60 to 67 million in 2050, which means that France will probably maintain its current share of 16 per cent. Population growth in France is only slightly lower than in the United Kingdom. The UK population is expected to increase from the current 59 million (15 per cent) to 68 million (16 per cent) in 2050. Italy is expected to loose more than one million inhabitants, from just over 57 million (15 per cent) today to 56 million (13 per cent) in 2050.

Population figures in 2050 very uncertain

In addition to predicting stronger population growth, the new forecast is also innovative because it quantifies demographic uncertainty. We have not only predicted the population as a number but as a complete probability distribution. The predicted number of 427 million inhabitants in 2050 is just one number in the distribution - the average value. We can interpret this as our best guess. However, since the future is uncertain, we cannot exclude the possibility that the population in EEA+ countries in 2050 will differ from 427 million. The forecast gives us probability estimates for such higher or lower numbers. For instance, the UN expects 400 million in 2050. The new forecast predicts with 77 per cent probability that there will be more than 400 million in 2050. In other words, the UN forecast is probably too low, but it is not impossible. In addition, there is an 83 per cent chance that the EEA+ population in 2050 will be larger than the current 392 million. Thus, although we expect an increase in the EEA+ population as a whole, a decline cannot be excluded, but it has a low probability.

We have summarised the probability distributions in the form of 80 per cent prediction intervals. Figure 1 "Population in EEA+ countries" shows that there is an 80 per cent probability for a population in EEA+ countries in 2050 of between 381 and 474 million. The margin is so wide because we cannot be very certain about demographic trends so far into the future. Prediction intervals for ten years ahead, for example, are much narrower. Among the three components that determine future population, international migration and the uncertainty attached to it strongly determines long-term uncertainty in population size. Uncertainty in the other two components, i.e. fertility and mortality, is somewhat less important. However these two components have an impact on the age structure of the population, both in the long and short term.

Ageing is certain

The new forecast confirms the message from earlier forecasts that there is hardly any doubt about ageing in the countries involved. At present, 17 per cent of the EEA+ population is aged 65 or older and this share is expected to increase to 29 per cent in 2050 (Figure 2; the age pyramid for EEA+ countries). For every 100 persons aged 20-64 in 2050, we can expect 57 persons aged 65 and over. This so-called Old Age Dependency Ratio (OADR) will be twice as high in 2050 as its current value of 28 per 100. The increase in the OADR is the main factor behind concerns about the sustainability of state pensions in many countries. In the future, the 65 and 65+ share will increase, and that of the 20 to 64 will fall. As early as 2010, it is almost one hundred per cent likely that the OADR value that year will be higher than the current 0.28, and this is no less likely for subsequent years. Although 0.57 is our best guess for the OADR in 2050, we cannot rule out a higher value: a value of at least 0.60 still has a probability of 23 per cent! At present, Ireland has the youngest age structure (11 per cent of the population is aged 65 or older) and Italy the oldest one (19 per cent). By the middle of the century, we can expect the lowest share of 65+ in the Netherlands (23 per cent), and the highest one in Spain (35 per cent). For the EEA+ as a whole, the share will be 29 per cent.

Lower mortality

There are three major reasons for more optimistic mortality assumptions. First, there is considerable room for improvement. Current life expectancies in the 18 countries range from a low 77.2 years in Belgium and Portugal, to high values of 79.9 years in Italy and Sweden, and 80.4 years in Iceland and Switzerland (Council of Europe 2003; Eurostat 2004). Japanese men and women have even higher life expectancies - the highest in the world at 81.8 years in 2002 (Japan Statistics Bureau 2005). Thus, the Japanese experience proves that higher life expectancies are a realistic possibility. Second, no matter how mortality is analysed (directly via mortality rates, via life expectancy, or via record life expectancy), all trends lead to higher life expectancies than assumed in past forecasts. For example, the record life expectancy has increased linearly by an average of 2.3 years per decade since 1840, when it was 44 years (Oeppen and Vaupel 2002). By record life expectancy, we mean the highest life expectancy observed internationally in a given year. There is no indication that the record life expectancy will be less, or stagnate, in the near future. Third, population forecasters have continuously underestimated the increases in future life expectancy. Ten-year life expectancy forecasts made by European statistical agencies after the Second World War were an average of 1.3 years too low. The difference was a total of 3.5 years for a forecast horizon of 20 years (Keilman and Pham 2004). For these reasons, we assumed relatively high life expectancy values in 2050: they range from 84.5 years in the Netherlands to 88 years in Spain. The UN assumes lower values: between 82 (Greece) and 86.1 years (Iceland); see UN (2005). Eurostat assumes life expectancies in 15 EU countries in 2050 that on average are 1.9 years (women) and 2.4 years (men) lower than ours.

Higher immigration

Future immigration to the EEA+ countries is likely to be higher than in the past. The gap in economic well-being between Europe and developing countries will continue to attract immigrants. Economic and political crises in the South and East will continue to put pressure on the wealthy EEA+ region. However, the ageing of European populations will make it necessary for the national governments not only to allow interested immigrants to come, but also that they actively seek new labour. For example, large numbers of elderly who need care will attract workers in the health sector from abroad. People would certainly be expected to come from the new EU member states, but also and primarily from outside Europe. As migrant populations are growing in EEA+ countries, family related migration is likely to increase in importance. However, business cycles and potential social tensions created by the transition of countries to a more multicultural society may lead to large fluctuations in migration flows, as has been the case in the past. However, the overall trend in net migration is upwards, as has been the case for the past 45 years (Figure 3 - net immigration to the EEA+ region), and the assumption is that it will continue for the forthcoming decades. The new forecast assumes that by 2050 the annual level of net migration into the EEA+ countries is 1.3 million. The linear trend observed since 1960 (with considerable fluctuations) would have led to two million (Keilman and Pham 2004).

Germany, Italy, and Spain will receive most immigrants

By 2050, large annual immigration flows are expected for Germany (nearly 290 000 in terms of net immigration), Italy (260 000), and Spain (almost 180 000). Relative to current population size, affluent Luxembourg can expect to receive most immigrants in 2050: six per thousand, well above the EEA+ average of 3.4 per thousand.

Norway

For Norway, a population of 5.75 million is forecast for 2050, slightly higher than Statistics Norway’s figure in its most recent official population forecast made in 2002. The latter forecast amounts to 5.59 million (forecast variant with medium population growth; see table .The 80 per cent forecast interval in 2050 stretches from 4.84 to 6.76 million. There is an 80 per cent probability that the share of the population aged 65 years and over in 2050 will be between 19.3 and 28.2 per cent. Thus, there is a great deal of uncertainty about this share, but it is 99 per cent certain that the share will be higher in 2050 than today (14.7 per cent).

Implications for welfare systems

This new forecast alerts us to the fact that we cannot foresee demographic developments accurately. This has major implications on how we plan pension systems and how we analyse future health care provisions. To recognise uncertainty this way leads to new criteria for evaluating policy reforms. In particular, a reform must be sufficiently robust so that unexpected demographic developments do not undermine its foundations or, say, put different cohorts in an inequitable position relative to taxes and benefits.

In the past, the volatility of demographics has often been underestimated. Demographic forecasters have been repeatedly surprised by large errors in historical forecasts. The magnitude of demographic uncertainty should make us careful when recommending policies to avoid unsustainable situations caused by ageing. For a risk-averse population, the cost of an outcome that is worse than expected outweighs the benefits of what is saved when the trend is better than expected. Sensible ageing policies should be especially robust in relation to a worse demographic future than that expected.

References :

Council of Europe (2003), Recent Demographic Developments in Europe 2003 (Strasbourg, Council of Europe Publishing).

Eurostat (2004), Population Statistics : 2004 Edition (Luxembourg, Office for Official Publications of the European Communities).

Japan Statistics Bureau (2005) Japan Statistical Yearbook 2005 (Tokyo, Statistics Bureau 2005).

Keilman, N., Pham, D.Q. (2004) Empirical errors and predicted errors in fertility, mortality and migration forecasts in the European Economic Area (Oslo, Statistics Norway). Available at http://www.ssb.no/publikasjoner/DP/pdf/dp386.pdf .

Oeppen, J., Vaupel, J. (2002) Broken limits to life expectancy. Science 296 , 1029-1031.

United Nations (2005) World Population Prospects : The 2004 Revision (New York, United Nations).

(1) Universitet i Joensuu, Finland.

(2) Statistics Netherlands.

(3) Netherlands Interdisciplinary. Demographic Institute.

(4) Statistisk sentralbyrå.

(5) Statistics Finland.