United Kingdom population trends in the 21st century

Transcription

1 United Kingdom population trends in the 21st century Chris Shaw, Government Actuary s Department (GAD) INTRODUCTION At the dawn of the 21st century, many of the probable key trends in population size and age structure over the next fifty years appear to be well known. In common with virtually all other developed countries, the United Kingdom has been experiencing historically low levels of fertility for many years. If this continues, and providing it is not offset by very high levels of net immigration, the population of the United Kingdom will peak in size sometime before 25 and then start to decline. The same prospect is in store for most other countries in the European Union 1 and for the developed world as a whole. 2 Population decline may bring both advantages and disadvantages 3 although problems associated with population size may depend more on the speed, rather than direction, of population change. The most significant population development over the next fifty years is likely to be a marked ageing of the population. The average age of the population will rise and the number and proportion of elderly people will increase. Largely, these changes are the result of long-term falls in birth and death rates made more complex by past fluctuations in births; in particular, the large cohorts born following the Second World War and during the 196s baby boom followed by the much smaller cohorts born since the early 197s. These low birth and death rates, and the older population structure which goes with them, are generally thought to be a permanent feature of populations in advanced societies. This article considers the likely trends in population size, age structure and potential support ratios in the United Kingdom over the coming century. The extent to which these trends could be modified by changes in fertility or migration levels is examined. The findings of a recent United Nations report on Replacement Migration are also considered in the light of the Government Actuary s official population projections for the United Kingdom together with new projection scenarios prepared specially for this article. Continuing increases in longevity, as predicted by most projection makers, would reinforce this ageing trend. The consequences of population ageing are far reaching (for pension provision, health and care services for the elderly, employment and training, saving and consumption patterns, housing, etc.) and continue to be widely discussed. 4,5 37 National Statistics

2 Population Trends 13 Spring 21 By considering the latest official population projections for the UK, as well as a number of additional projection scenarios (see Box one), this article examines whether these long-term population trends are inevitable and the extent to which they could realistically be altered by changes in demographic behaviour. A recent United Nations report 6 has focussed attention on the possible role of migration in this context. (It should be noted that although the UN report specifically considered Box one PROJECTIONS, FORECASTS, VARIANTS AND SCENARIOS National population projections are undertaken by most industrialised countries and by international organisations such as the European Union and the United Nations. A population projection shows what would happen if particular assumptions regarding future fertility, mortality and net migration were to occur in practice. However, to give users an indication of the inherent uncertainty in demographic behaviour, an organisation will often produce a range of projections based on alternative assumptions, usually known as variant projections. Normally, one of these will be identified as the principal, medium or central variant. Strictly, such a projection is still simply just the outcome of a particular set of assumptions, but it is inevitable that users will treat it as being a forecast of the most likely course of future events. The official UK national population projections carried out by the Government Actuary s Department and the regular revisions of world population projections produced by the United Nations fall into this category. Although inevitably subject to error, these projections provide a vital guide to short and long-term planning in many different fields. In addition, a scenario or what if projection may sometimes be made to illustrate the consequences of a particular, but not necessarily realistic, set of assumptions. The UN replacement migration scenarios, the stationary population projection and others discussed in this article, are examples of this type of projection. Finally, it is worth noting that in some of the scenario projections discussed in this article, the usual process of projection making has been reversed. Normally, a set of assumptions is specified which determines the population outcome. However, in some of the scenarios considered in this article, the desired population outcome is specified and the assumptions required to produce this outcome are then calculated. UNCERTAINTY OF LONG-TERM PROJECTIONS In reporting the results of GAD s national population projections for the UK, attention is normally focussed on the first twenty-five years. However, the official projections for the United Kingdom are prepared for a period of seventy years from their base date. For the purposes of this article, as it is concerned specifically with long-term population trends, all the projections have been extended to the year 21. Of course, as can be noted in particular from Figures 4 and 5, uncertainty increases appreciably with time and the range of possible outcomes in one hundred years time is vast. increases in immigration, increases in net migration could also be achieved by reductions in emigration.) Coinciding with this work, there has recently been considerable public debate in western countries about the pros and cons of encouraging immigration with a view, in particular, to tackling skills shortages in the job market. The UK government, itself, last year called for a debate to begin on the benefits and challenges of managed migration. 7 However, although the two are related, there is a distinction to be drawn between migration as a possible solution to skills shortages (a specific, more short-term issue) and migration as a potential remedy for long-term population ageing. It is the latter issue with which this article is concerned. In this article fertility change, as well as migration, will be considered. Whether it would be possible to bring about substantial changes in either fertility or migration levels, even if it were considered desirable, is another matter entirely. Most researchers have tended to conclude, for example, that family friendly economic measures tend to have rather modest effects on fertility levels. 8 And the scope for influencing migration levels and flows - which depend, amongst other things, on the socio-economic and political conditions in both receiving and sending countries - is clearly limited. Although different mortality levels would also, of course, have an impact on future population size and age structure, this is largely ignored in this article. It is assumed that higher mortality levels are unlikely to become a popular solution to the problems of population ageing! Two indicators of population size are considered in this article: the total population and the population of working age. For the purposes of this article, working age is taken to be 15 to 64, for ease of comparison with the recent UN work 6 and for consistency through time. Normally, in reporting the results of the official UK national projections, working age is taken to be from 16 to state retirement age. Between 21 and 22, state retirement age in the UK will change from 65 years for men and 6 years for women, to 65 years for both sexes. 9 But in this article, the 15 to 64 definition is used throughout. Two indicators of population ageing (median age and the proportion of the population aged 65 and over) are also considered, as is the elderly Figure 1 Population (millions) Actual and projected population by broad age group, United Kingdom, National Statistics 38

3 demographic support ratio - for the purposes of this article defined as the ratio of persons aged 15 to 64 to those aged 65 and over. It should be emphasised, however, that demographically defined potential support ratios, whatever age boundaries are used, take no account of workforce participation rates and therefore do not represent the real level of economic dependence. Projections of the contributory support ratio in Great Britain - the ratio of contributors to the National Insurance Fund, to those in receipt of pensions - are published elsewhere. 1 PAST AND FUTURE TRENDS IN KEY DEMOGRAPHIC VARIABLES As these projections look forward a century, it is instructive to begin by looking back a similar period at the changes which occurred in the 2th century in population size and age structure in the United Kingdom. Figure 1 shows actual trends since 191 together with projected future trends taken from GAD s latest (1998-based) principal projection, 11 but extended to 21. In the context of a population that has increased from under 4 million in 191 to just under 6 million in 1999, a projected further increase to 65 million in 236 and subsequent slow decline represents comparatively modest change. However, the pace of population ageing is set to accelerate markedly from around 21 when the large cohorts of people born after the Second World War begin to reach age 65. Nevertheless, the UK has already seen its share of elderly people in the population rise from 5 per cent in 191 to 15 per cent now. Conversely, the proportion of people aged to 14 has fallen from 33 per cent to 19 per cent over the same period. Figure 2 Key population indicators, principal and stationary projections, United Kingdom, Population (millions) Total population Principal projection Stationary projection Population (millions) Working age population Principal projection Stationary projection Percentage 25 Percentage over 65 Principal projection Ratio 5. Support ratio (15 to 64/65 and over) 2 15 Stationary projection Stationary projection Principal projection. Age 45 Median age Principal projection Long-term assumptions 4 35 Stationary projection Children Annual per net inward woman migration Mortality Principal , improving 3 Stationary 2.75 constant 39 National Statistics

4 Population Trends 13 Spring 21 GAD PRINCIPAL PROJECTION COMPARED WITH A STATIONARY POPULATION PROJECTION The size of the future population and its age structure is not just determined by future levels of fertility, mortality and migration, but also by the age distribution of the population alive today. This is shown by Figure 2, which compares GAD s 1998-based principal projection with a stationary population projection. The official principal projection assumes a long-term average family size of 1.8 children per woman (well below replacement level - see Box two), continuing improvements in mortality and long-term net inward migration of 95 thousand a year. 12 Figure 2 shows how the different population indicators are projected to develop under the principal projection assumptions, and compares them to what would happen if there were to be an abrupt shift to the conditions that would lead to a stationary population. A stationary population has an unchanging size and age structure and arises, eventually, given replacement level fertility, constant mortality rates at all ages and zero net migration at all ages. These are, of course, purely hypothetical and extremely unrealistic conditions. There is no immediate likelihood of fertility returning to replacement level, mortality improvement ending or net migration returning to zero, let alone all three happening simultaneously. However, it is useful to consider the extent to which the population would not be stationary under these conditions in the short and medium-term as this reflects the inherent population momentum resulting from the existing population age structure. Clearly, the stationary projection assumptions are very different to those underlying the latest official principal projection. The principal projection assumptions of inward migration and improving mortality produce a larger population than the stationary projection although this is gradually offset by the different fertility assumptions. The improving mortality and lower fertility assumptions of the principal projection also lead to an older population than in the stationary projection. Box two REPLACEMENT LEVEL FERTILITY, AVERAGE FAMILY SIZE AND THE TOTAL FERTILITY RATE Replacement level is the level of fertility required for the population to replace itself in size given prevailing mortality rates and in the absence of migration. Replacement level is now around 2.75 in the United Kingdom, i.e. women would need to have, on average, 2.75 children each to ensure the long-term natural replacement of the population. In fact, average family size has been falling steadily from a peak of nearly 2.5 children per woman for women born in the mid 193s. In the latest GAD principal projection, it is assumed to level off at 1.8 children per woman for women born in or after Average family size is a cohort measure, measuring fertility according to a woman s year of birth. The analogous period statistic is the total fertility rate (TFR). This is a hypothetical measure giving the average number of children who would be born per woman, if women were to experience the age-specific fertility rates of the year in question throughout their childbearing lives. Therefore, if age-specific fertility rates are assumed to remain constant from some point in the future, as is normally the case in population projections, then the long-term TFR and the long-term average family size will be the same. Figure 3 Annual births, United Kingdom, Births (thousands) 1,2 1, Nevertheless, although the levels of the projections differ, the timing of turning points and directions of population change are broadly similar. This demonstrates the importance of the existing population structure in determining future population trends. In particular, the relative sizes of birth cohorts born between the Second World War and the mid 197s (see Figure 3) will continue to be a powerful influence on the UK s population size and age structure into the second half of the 21st century. The annual number of births in the UK rose quickly from around 8 thousand in 195 until it reached a peak of over one million in The total then fell rapidly to a post war minimum of around 65 thousand in Since 198 the annual number of births has been much more stable. However, as Figure 2 shows, even assuming an immediate move to stationary population conditions, the large fluctuations in the size of the pre-198s birth cohorts would affect the size of the working age population until they have all reached age 65 (in the 24s) and the size of the total population for many further years. So, even with an immediate shift to stationary population conditions, it would take over half a century to produce a stationary population. This indicates the difficulty, perhaps even the impossibility, of effecting short-term solutions to perceived population problems by demographic means. It might initially appear a little surprising that the percentage of the population aged over 65 and the potential support ratio take longer to level off in the stationary projection than in the principal projection. In the principal projection, these indicators stabilise around 24 once the ups and downs in the sizes of pre-198s birth cohorts have worked their way through to elderly ages. However, in the (unrealistic) stationary projection, the assumption of an immediate move to replacement level fertility greatly increases the size of post-2 birth cohorts. In time this increases the size of the working age population, and so after 24 the percentage of elderly people reduces and the potential support ratio rises until these larger cohorts start to reach 65 in the 26s. UNITED NATIONS REPLACEMENT MIGRATION REPORT As noted above, a recent United Nations report 6 has focussed attention on the extent to which different levels of net migration could alter future population size and alleviate population ageing. The UN report considered the numbers of migrants required to achieve particular population outcomes; namely keeping the total population constant, keeping the working age population constant, and keeping the working age to post working age support ratio constant. Projections were calculated for eight major countries, including the United Kingdom, for each of whom the UN projects long-term fertility to remain below replacement level. 13 National Statistics 4

5 Table 1 United Nations Replacement Migration scenarios: summary of United Kingdom results Scenario I II III IV V Official UN Constant Constant Constant medium Zero total ratio variant migration population population 15 to 64/65+ Total population (1995 = 58.3 million*) Population aged 15 to 64 (1995 = 37.8 million*) Potential support ratio 15 to 64/65+ (1995 = 4.1*) Average annual net inward migration in given scenario (thousands) ,441 * UN estimates of the population at mid These differ slightly from official ONS estimates. Table 1 summarises the scenarios considered and compares them with the latest official UN medium variant projection 2 which is shown as Scenario I. 14 All scenarios use the medium variant fertility and mortality assumptions. In Scenario II, the medium variant migration assumption is replaced by an assumption of zero future net migration. From this scenario, the UN obtained the maximum total population size, maximum working age population and maximum support ratio that would result in the absence of future net migration, and the year in which these maxima would be reached. They then calculated the level of net migration required in later years to maintain total population size (Scenario III), working age population (Scenario IV) or potential support ratio (Scenario V) at these maximum levels. In the case of the United Kingdom, the UN calculated that with zero net migration, the total population of the UK would peak at 58.8 million at 22 and the working age population would peak at 38.9 million at 21. Maintaining the total population size at 58.8 million until 25 would require average annual net inflows of around 1 thousand persons in the second quarter of this century, while holding the working age population at 38.9 million would require an average inflow of 13 thousand migrants a year during this period. However, the scenario which attracted most attention was the one where the support ratio of those aged 15 to 64 to those aged 65 and over was held constant. For the UK, the maximum value of the potential support ratio under the UN s zero net migration scenario was its initial (1995) value of 4.1. Thereafter, the ratio falls continually until 24. Indeed, as Figure 7 below shows (based on the GAD projections), the support ratio is set to decline substantially under any plausible set of future assumptions. So, maintaining the support ratio at the current level is a wholly unrealistic scenario. Nevertheless, the UN s calculation that the UK would require an average of over one million migrants a year to maintain this maximum ratio until 25 (more than doubling the total population in the process) has generated considerable media comment. The absurdity of this scenario is demonstrated by the UN s calculation that to keep its potential support ratio at current levels, the Republic of Korea would require net inward migration of around 1 million every year. This would raise its population at 25 to over 6.2 billion, larger than the entire population of the world today. It is unfortunate perhaps that this scenario was presented on an equal footing to the others in the UN report and has deflected attention from some of their more useful findings. Although the annual averages shown in Table 1 for Scenarios III and IV may not seem implausible, they conceal considerable fluctuations within the time period. As noted above, the UN first calculated the maximum population size (or support ratio) that would occur in the absence of net migration. In the case of the UK, the UN found that the maximum total population size, given zero net migration, would occur in 22. Thereafter, net migration would need to increase quickly to maintain this total population size. Indeed, the UN calculated that annual net migration would need to exceed 1 thousand a year from 23 onwards in this scenario. These fluctuations become larger and more erratic when the working age population is considered. The problem here is that all members of a given birth cohort reach the working age boundaries (here defined as 15 and 65) at the same time. So, in attempting to keep the working age population constant, net migration would effectively have to compensate for irregularities in the size of birth cohorts. In fact, in Scenario IV, the required net inflows vary from zero initially to nearly 3 thousand per year during The problem is greater still when attempts are made to hold the potential support ratio constant. To maintain a constant support ratio of four, say, an increase of 4, in the working age population would be required for every 1, increase in the population aged over 65. As migrants tend to be highly concentrated at young ages (see Box three), the required peaks and troughs in the migrant flow would lead to exploding cyclical net migration rates. 15 Box three MIGRANT AGE DISTRIBUTIONS Typically, migrant age distributions are highly peaked at young working ages. In GAD s official national population projections, age distributions derived from recent International Passenger Survey data are applied. These distributions are heavily peaked around age 2. These age distributions have been used in all the special projections run by GAD for this article. The UN applied an age distribution derived from data from Australia, Canada and the USA, again highly peaked at young working ages, for their replacement migration scenarios. Research by GAD has shown that age distributions remain very stable from year to year. 16 Nevertheless, it should be noted that alternative age distribution assumptions could produce very different projection outcomes. For example, a significant net migration outflow at retirement ages would act to improve support ratios. In practice, however, migration flows at older ages tend to be very small. To illustrate this point, GAD ran a projection to calculate the annual net migration figures needed to hold potential support ratios constant from year to year at current levels. This produced figures which varied from net inflows of over 1.5 million a year in the late 22s when the large baby boom cohorts of the mid 196s will reach age 65, to some net outflows in the early 24s when the small cohorts of the late 197s leave the working age group. Extending the projection to 21 gives annual net inflows which, by the end of the century, vary from 1 thousand to 5 million within a ten year period. Quite aside from the considerable consequences for total population size, it would clearly be impossible to fine tune annual migration figures in such a way. 41 National Statistics

6 Population Trends 13 Spring 21 Furthermore, maintaining broadly constant annual numbers (a somewhat less implausible aim) would not produce the same results. For example in their extreme Scenario V, the UN reported that the UK would require an annual average of around one million net migrants over the period to 25 to maintain potential support ratios at current levels. But a GAD projection assuming a constant net inflow of one million migrants a year sees the support ratio improving from 4.15 in 1998 to reach a peak of 5. in 211, before worsening steadily to around 3.5 by 25 and 3. by 21. This is shown in Figure 8a and is discussed in more detail below. The UN s report focussed solely on migration. However, similar calculations could have been made of the fertility levels needed to maintain particular population sizes or support ratios although, of course, there would be a fifteen year time lag before higher levels of births could have any effect on the size of the population of working age as defined in this article (ages 15 to 64). Such computations would inevitably result in similar annual fluctuations to those discussed above for migration. As noted above, the latest (1998-based) principal projection for the United Kingdom is based on an assumption of a long-term family size of 1.8 children per woman and long-term net inward migration of 95, persons a year. In GAD s official variant population projections, 11 high and low fertility assumptions of 2. and 1.6 children per woman are used, i.e. ±.2 children compared with the principal projection. For migration, the high and low long-term assumptions are for net inflows of 145 and 45 thousand each year, i.e. ± 5 thousand compared with the principal projection. These are intended as plausible alternative scenarios and not to represent upper or lower limits for future demographic behaviour. Combining the three alternative fertility assumptions (high, principal and low) with those for migration therefore gives nine alternative projections (see Box four) encompassing a broad, but still plausible, range of future population outcomes. These nine projections provide the starting point for the further discussion in this article. For example, to investigate the possibility of achieving stability in the key population indicators in the second half of this century, we will take whichever is the most suitable of these nine projections to the year 25 and then consider the changes required to either the total fertility rate (TFR - see Box two) or annual net migration, assuming the other remains fixed, in order to hold the particular indicator constant. Clearly, this is a very crude approach. A major one-off shift in fertility or migration at 25 preceded and succeeded by constant levels is not a realistic prospect. And the examples given will not be unique solutions. The same total population size at 21, for example, could be brought about by many different combinations of fertility and migration (and indeed mortality) levels. The intention is only to give a broad indication of the likelihood of the particular population outcome being achieved. ALTERNATIVE ASSUMPTIONS OF FUTURE FERTILITY AND MIGRATION In the remainder of this article, we will follow the practice of earlier demographers, 15 and abandon any attempt to fine tune population size or age structure on an annual basis. Instead, we will consider how our key indicators of population size and ageing could be affected by alternative constant long-term assumptions of future fertility and migration. As in the UN work, the simplifying assumption is made that different net migration levels will not change overall fertility and mortality rates. Box four SUMMARY OF ASSUMPTIONS UNDERLYING OFFICIAL FERTILITY AND MIGRATION VARIANT PROJECTIONS Fertility Annual net inward migration High Principal Low Projection label HF HM HF PM HF LM High TFR Net migration (s) Projection label PF HM PP PF LM Principal TFR Net migration (s) Projection label LF HM LF PM LF LM Low TFR Net migration (s) TOTAL POPULATION SIZE At present, the population of the United Kingdom is growing both through natural increase (more births than deaths) and through net inward migration. As births currently exceed deaths by nearly 1 thousand a year and with little to suggest a reversal in net migration trends, population growth is set to continue for some time as illustrated by Figure 4. However, in the principal projection for the UK, deaths begin to outnumber births in about thirty years time. By 236, the growing natural decrease exceeds the assumed continuing gain to the population through net migration and so the population slowly begins to decline in size. It is clear from Figure 4, however, that population decline is not inevitable. For example, a long-term TFR of 2., coupled with net migration at or above the level assumed in the principal projection, would produce continued population growth throughout the 21st century. Under the high fertility, high migration variant HF HM, the population would be growing by around one quarter of one per cent per year at the end of the century. The UN Replacement Migration report considered the levels of net migration needed to maintain total population size at a constant level. It Figure 4 Millions Total population of the United Kingdom under principal and variant projections, [See Box four for explanation of projection labels] PF HM HF LM HF HM HF PM PP LF HM PF LM LF PM LF LM National Statistics 42

7 Figure 5 Millions Population aged under principal and variant projections, United Kingdom, [See Box four for explanation of projection labels] PF HM HF LM HF HM HF PM is not self evident that constant population size is a desirable aim although, as noted above, there may well be problems associated with rapid change - in either direction. However, it is interesting to note that the high migration variant PF HM (where net inflows of 145 thousand a year effectively compensate for a below replacement level TFR of 1.8) would produce a very stable population size in the second half of the century. However, at 67 million, this would be considerably greater than the population size today. In the context of a population that rose from under 4 million in 191 to nearly 6 million in 1999 (Figure 1 above), it could be argued that the principal projection itself (rising to 65 million by 236, but returning to 6 million by 21) represents a fairly stable future. However, the three low fertility projections and the low migration variant PF LM shown in Figure 4 would all lead to declining populations by the end of the century. Under the low fertility, low PP LF HM PF LM LF PM LF LM migration variant LF LM, the population would be falling by about one half of one per cent per year by the year 21. It is, therefore, not likely that the population size could stabilise at as low a level as, say, 6 million, within this time frame. For example, the low migration variant PF LM would produce a population of around 6 million in 25. But further GAD work shows the TFR would then need to rise from 1.8 to nearly 2.1 or annual net inward migration from 45 thousand to around 125 thousand (or some equivalent combination of higher fertility and migration) to keep the population around 6 million for the remainder of the century. And stabilisation at lower levels would be even more improbable. Essentially, the levels of fertility or migration required to bring the population down below 6 million by 25 would be incompatible with the levels required to maintain that level thereafter. WORKING AGE POPULATION The outlook for the working age population is shown in Figure 5. All the fertility and migration scenarios considered show some further increase in the total number aged 15 to 64. But in most cases this would come to an end around 211 when the large cohorts born following the Second World War start to reach age 65. [If the conventional definition of working age 16 to state retirement age had been used, then the peak would be delayed to 22 when the phased transition of women s retirement age from 6 to 65 is complete.] In the principal projection, the working age population peaks at 41 million, and would fall to 38.5 million by 25 and 36 million by 21. However, as with the total population, long-term decline is not inevitable and the working age population would continue to increase throughout the century under the high fertility scenarios HF HM and HF PM. Again, the high migration variant (PF HM) would produce a fairly stable working age population at just over 4 million in the second half of the century. To hold the working age population constant at significantly higher or lower levels would require sizeable shifts in fertility or migration or both. For example, the high fertility variant HF PM would lead to a population approaching 42 million by 25. But to maintain the group at this size, further GAD work shows the TFR would then need to fall from 2. to just above 1.8, or annual net inward migration would need to fall from 95 thousand to 4 Figure 6 Population ageing under principal and variant projections, United Kingdom, Age (years) Median age LF LM LF PM LF HM PF LM PP PF HM HF LM HF PM HF HM Percentages Proportion of the population aged 65 and over LF LM LF PM LF HM PF LM PP PF HM HF LM HF PM HF HM [See Box four for explanation of projection labels] National Statistics

8 Population Trends 13 Spring 21 thousand (or some equivalent combination). Conversely, the principal projection PP would see the working age population decline to nearly 38 million by 25. But to maintain this level, the TFR would then need to rise from 1.8 to 1.95, or annual net inward migration from 95 thousand to 135 thousand. AGEING POPULATION Whereas Figures 4 and 5 showed that population decline is not inevitable, Figure 6 demonstrates that population ageing will occur under any plausible set of future assumptions about future fertility or migration. Whether measured by median age or the proportion of people aged over 65, the population is going to grow rapidly older in the first half of this century. Of course, ageing also depends crucially on the future level of mortality. But population ageing would take place even if there were to be no further improvement in life expectancy. This was illustrated by the stationary projection in Figure 2 where even with mortality rates held constant (and fertility at replacement level), median age and the proportion of people aged over 65 would rise substantially over the next forty years. Although the results are obviously dependent on the assumptions chosen, Figure 6 clearly suggests that the extent of future population ageing is much more sensitive to likely variations in future fertility than to future net migration. Under the three high fertility scenarios (TFR = 2.), median age would still rise from 37 years now to around 41 years in the long-term. However, under the low fertility scenarios (TFR = 1.6), there would be a far greater increase to around 47 years. In contrast, the very different levels of net migration have comparatively little effect on median age. Compared with the low migration variant, the high migration variant (an extra 1 thousand migrants each year) would only reduce the median age at 21 by between half a year and one year. Similar results are seen for the proportion of the population aged over 65. Although higher fertility levels would reduce population ageing, even the high fertility, high migration scenario HF HM would see the median age rising by three years between 2 and 22. This is due to current age structure, which in turn is a result of changes in the past numbers of births (Figure 3). Thus, during the first half of this century, the number of elderly people will rise as the relatively large cohorts Figure 7 Ratio of persons aged to those aged 65 and over under principal and variant projections, United Kingdom, Ratio [See Box four for explanation of projection labels]. HF HM HF PM HF LM PF HM PP PF LM LF HM LF PM LF LM born after the Second World War and during the 196s baby boom replace at older ages the much smaller cohorts born before Conversely at younger ages, the relatively small post-1975 cohorts will replace the baby boomers. Under all these scenarios, median age and the proportion of the population aged over 65 level off and remain fairly constant from around 25. In practice, long-term trends will not be so smooth, but will reflect whatever fluctuations occur in the size of future birth cohorts. These erratic annual fluctuations cannot be predicted now but, in practice, even though the major changes between 1945 and 1975 may not be repeated, the trend in future annual births is likely to be far from smooth. POTENTIAL SUPPORT RATIOS The combination of a stable or declining (in most scenarios) working age population and sharply rising elderly population will cause the elderly demographic support ratio to fall dramatically, as shown in Figure 7. In 1999, there were 4.2 persons aged 15 to 64 for every person aged 65 and over. This will fall steadily until around 24, although there would then be some modest recovery under the high fertility scenarios. Ultimately, the potential support ratio would level off around 2.75 (with a TFR of 2.), 2.45 (TFR = 1.8) or 2.15 (TFR = 1.6). Again, changes in the migration assumptions have a relatively minor effect when compared with changes in the fertility assumptions. The inefficiency and ineffectiveness of net migration as a tool for reducing support ratios is further demonstrated by Figure 8a. This charts not just the principal projection, but also projection scenarios assuming respectively annual net inward migration of zero,.25 million,.5 million and one million (each with the principal projection assumptions of fertility and mortality). The highest assumptions are, of course, quite unrealistic. The UK has never experienced net inward migration of over 2 thousand persons in any year, although record net inflows of around 18 thousand have been recorded in both 1998 and But the very high migration scenarios clearly demonstrate that, although they would provide a short-term boost to support ratios, even massive constant net migration inflows would not prevent support ratios ultimately falling. The simple reason why even large constant net migration inflows would not prevent support ratios from falling in the long-term is that migrants grow old as well! Although, some immigrants to the UK will leave the country before reaching retirement age, it is assumed implicitly from the age distributions used in this work (see Box three) that most will remain in the population. So, although a steady large inflow of young migrants would continue to boost the working age population, before long it would also start adding to the retirement age population and a four to one (say) potential support ratio could not be maintained. Figure 8b shows the effect that different long-term fertility levels would have on the potential support ratio. Interestingly, a long-term TFR of 2. children per woman would produce much the same support ratio at 21 as would annual net migration of half a million people a year, but with a total population of 75 million rather than 12 million. And unlike migration, significantly higher fertility levels, if they could be achieved, would produce markedly higher long-term support ratios. Although, of course, higher fertility would also increase the number of child dependants. There are other ways of altering the ratio of those in the working ages to those in retirement. For convenience, in this article, 65 has been used as the upper boundary for working age. Figure 8c shows how higher upper boundaries of working age would produce higher support ratios. In fact, under the principal projection assumptions used for this graph, an increase of over seven years in the upper boundary of working age (to 72.3) would be needed to produce the same support ratio in the National Statistics 44

9 second half of this century as in Of course, in the UK, actual state retirement age is currently 65 years for men but only 6 years for women, although between 21 and 22 it will change to 65 years for both sexes. In the normal reporting of national population projections, state retirement age is used to mark the boundary between working age and retirement. Defined in this way, the forthcoming change in women s state retirement age will have a considerable effect on dependency ratios. 11 However, any attempt to the define the working age population by age boundaries is somewhat arbitrary. In reality, retirement starts at a range of different ages. Further, altering fertility or migration levels, or changing retirement age, are not the only ways of changing the number of people at working age. Indeed, research shows that changes in workforce participation rates have in the past been a more important factor than demographic trends in influencing real (economic) dependency. 18 And in practice, the average age at retirement in the UK is presently well below state retirement age. Comprehensive data is not available, but a recent survey of occupational pension schemes showed the mean age at retirement for members to be currently well below Indeed, whether through deliberate choice or otherwise, a third of people aged between 5 and state pension age are now not working. 2 Figure 8 Ratio Support ratio under alternative long-term assumptions, United Kingdom, (a) alternative migration assumptions TFR: 1.8 Annual net inward migration: as shown 1 million 1 / 2 million 1 / 4 million 95, (PP) Zero SUMMARY GAD s latest principal population projection suggests that the population of the United Kingdom will peak in size in the year 236. In common with all other developed countries, there will also be rapid population ageing largely as a result of past trends in birth and death rates, but reinforced by expected continuing improvements in mortality. Although eventual population decline now seems the most likely future course, it is not inevitable. A significant, but not unthinkable, rise in fertility could produce continuing population growth throughout the 21st century. Rapid population ageing, on the other hand, will occur under any plausible set of future assumptions. An unavoidable consequence of this is that the support ratio of people of working age to those above working age will fall rapidly. In theory, the UK s population size could stabilise in the second half of this century given appropriate fertility and/or migration levels. But it is not clear that this is a desirable aim and, in any case, the available evidence suggests that society s ability to influence demographic behaviour is rather limited. Despite much recent attention being focussed on migration, it is clear that this is not a long-term solution to the problems of population ageing. The future proportion of people aged over 65, for example, is much more dependent on future fertility levels than on migration. Similarly, higher fertility levels, if they could be achieved, could do far more to improve long-term support ratios than migration. However, measures such as raising workforce participation rates or discouraging early retirement are likely to remain a more practical tool for increasing the working population than attempting to influence demographic behaviour.. (b) alternative fertility assumptions Ratio 5.5 TFR: as shown 5. Annual net inward migration: 95, Ratio (c) alternative upper boundaries for working age (HF-PM) 1.8 (PP) 1.6 (LF-PM) TFR: 1.8 Annual net inward migration: 95, ACKNOWLEDGEMENTS Some of the projections prepared for this article have already been discussed in a paper by David Coleman from the University of Oxford. 3 The author is grateful to Dr Coleman for his helpful input to the present article. The author is also grateful to Adam Michaels of the Government Actuary s Department for his preparation of, and advice on, the numerous population projections specially run for this article / / / / / / / / National Statistics

10 Population Trends 13 Spring 21 Key findings Trends in future population size and age structure are not solely dependent on future levels of fertility, mortality and migration. There is considerable inherent population momentum resulting from the present age structure of the population. In particular, the fluctuating size of birth cohorts born between 1945 and 1975 will be a powerful influence on the UK s population size and age composition into the second half of this century. Although the total UK population is likely to peak in size sometime before 25, population decline is not inevitable. In particular, increases in fertility could lead to continued population growth throughout the 21st century. And in theory, given appropriate levels of fertility or migration, population size could stabilise in the second half of the century, but only at a somewhat higher level than today. The population of working age is likely to peak in size in around ten years time. But, as with the total population, long-term decline is not inevitable. However, the population will age, and consequently potential support ratios will fall, under any plausible set of future assumptions. In practice, even very large net migration inflows will not overcome population ageing or prevent support ratios from falling. Indeed, higher fertility levels, if they could be achieved, would have a more significant effect. REFERENCES 1. Shaw C, Cruijsen J, De Beer J and De Jong A. Latest population projections for the European Union. Population Trends 9. The Stationery Office (1997) pp World Population Prospects: The 1998 Revision. United Nations (1999). 3. *Coleman D. Who s afraid of low support ratios? A UK response to the UN Population Division report on Replacement Migration. From United Nations Expert Group Meeting on Policy Responses to Population Ageing and Population Decline. United Nations (2). 4. Dunnell K. Policy responses to population ageing and population decline in the United Kingdom. Population Trends 13. The Stationery Office (21) pp Foresight Ageing Population Panel. The Age Shift - A consultation document. Department of Trade and Industry (2). 6. *Replacement Migration: Is it a solution to declining and ageing populations? United Nations (2). 7. UK migration in a global economy. Speech by Barbara Roche MP, Immigration Minister. Home Office (September 2). 8. Teitelbaum M. Sustained below replacement fertility: realities and responses. From Below Replacement Fertility. Population Bulletin of the United Nations. Special Issues 4/41 (1999). 9. Pensions Act 1995 Chapter 26. Part II Section 126 and Schedule National Insurance Fund Long Term Financial Estimates, Chapter 4. Government Actuary s Quinquennial Review. Cm 446. The Stationery Office (1999) based national population projections. ONS Series PP2 no. 22. The Stationery Office (2). nsbase/onlineproducts/default.asp#population 12. Shaw C based national population projections for the United Kingdom and constituent countries. Population Trends 99. The Stationery Office (2) pp The other countries were France, Germany, Italy, Japan, the Republic of Korea, the Russian Federation and the United States. Projections were also calculated for Europe and the European Union as aggregates. 14. There are a number of differences between the Government Actuary s official principal projection and the UN medium variant projection for the United Kingdom. As well as different assumptions of future fertility, mortality and migration, the base population estimates used by the UN are out of line with the official estimates made by the Office for National Statistics. These differences are discussed in more detail by Coleman (reference 3 above). However, GAD has established that these differences would not have significantly altered the findings in the UN replacement migration report. 15. *Lesthaeghe R. Europe s demographic issues: fertility, household formation and replacement migration. From United Nations Expert Group Meeting on Policy Responses to Population Ageing and Population Decline. United Nations (2). 16. National population projections: a new methodology for determining migration assumptions. Occasional Paper 42. OPCS (1993). 17. International migration, National Statistics First Release (October 2). 18. Johnson P and Falkingham J. Ageing and economic welfare. Sage publications (1992). 19. Income Data Services. Early retirement patterns persist. From IDS Pensions Bulletin 128 (September 1999). 2. Winning the generation game: improving opportunities for people aged 5 65 in work and community activity. Cabinet Office (2). * All asterisked references are currently available on the United Nations Population Division web site popdecline.htm National Statistics 46