National population projections by age and sex are produced for the UK and constituent countries every two years. These sets of projections are based on a review of the underlying assumptions regarding fertility, mortality and migration.
The primary purpose of the projections is to provide an estimate of future population which is used as a common framework for national planning in a number of different fields. The main focus of the 2010-based projections is on the next 25 years up to 2035. Longer term projections to 2110 are also produced but these should be used with caution as population projections become increasingly uncertain the further they are carried forward, and particularly so for smaller geographical areas. The projections are being made publicly available to 2110 under the government's transparency agenda. In addition to the principal (main) projections, variant projections are also available, based on alternative assumptions of future fertility, mortality and migration. For more information on how ONS population projections meet users needs along with information on their fitness for purpose, please see the report on quality and methodology (290.9 Kb Pdf) on the ONS website.
The 2010-based projections replace the 2008-based projections published on 21 October 2009.
The projections were produced by ONS on behalf of the National Statistician and the Registrars General of Scotland and Northern Ireland. The underlying assumptions were agreed in liaison with the devolved administrations – Welsh Government, National Records of Scotland (NRS) and Northern Ireland Statistics and Research Agency (NISRA) - following consultation with key users of projections in each country and advice from an expert academic advisory panel.
The Government Actuary’s Department (GAD) was responsible for the production of the official national population projections for the UK and its constituent countries from 1954 to 2006. Responsibility for the production transferred to ONS on 31 January 2006. Results of previous sets of projections are available on the GAD website.
This report contains background information for the 2010-based national population projections. Included are sections on:
defining the base population,
the method of projection,
the projection period,
background on principal and variant projections,
the relationship of UK level projections to individual country level projections,
changes to the state pension age,
national population projections expert advisory panel,
summary of the long-term assumptions of future levels of fertility, mortality and migration.
The projections are based on estimates of the resident population of the UK and its constituent countries at mid-2010. The population includes all usually resident persons, whatever their nationality. Members of HM Armed Forces in the UK are included, but members of HM Armed Forces and their families who are abroad are excluded. Members of foreign armed forces in the UK are included, with any accompanying dependants.
The projections for England and Wales are based on the mid-2010 population estimates published by ONS on 30 June 2011. The projections for Scotland are based on the mid-2010 population estimates published by NRS on 27 April 2011, and likewise the projections for Northern Ireland are based on the mid-2010 estimates published by NISRA on 30 June 2011.
Official mid-year population estimates produced by ONS, NRS and NISRA are prepared by individual age to the age of 89, with an upper age band for all those aged 90 and over. Estimates of the population aged 90 to 109 and 110 and over are prepared using the Kannisto Thatcher survivor ratio method, with the results controlled to agree with the official estimates of all those aged 90 and over.
The projections are made for successive years running from one mid-year to the next. For each age, the starting population plus net inward migrants less the number of deaths produces the number in the population, one year older, at the end of the year. To this has to be added survivors of those born during the year. Age is defined as completed years at the last birthday.
Migration is assumed to occur evenly throughout the year. For computing purposes, this is equivalent to assuming that half the migrants in a given year at a given age migrate at the beginning of the year and half at the end of the year. The number of net migrants to be added to obtain the population aged x+1 at the end of the projection year therefore consists of half of those migrating during the year at age x and half of those migrating during the year at age x+1.
The number of deaths in a year is obtained by adding half of the net inward migrants at each age to the number in the population at the beginning of the year and applying the mortality rate qx. The mortality rates used in the projections represent the probabilities of death between one mid-year and the next, according to a person's age last birthday at the beginning of the period. The appropriate rate of ‘infant mortality’, that is the probability of a new-born child not surviving until the following mid-year, is also given. This is about 85 per cent of the full, first year of life infant mortality rate used in official statistics.
The number of births in the year is calculated by multiplying the average number of women at each single year of age during the year (taken as the mean of the populations at that age at the beginning and end of the year) by the fertility rate applicable to them during that year. The total number of births in a year is assumed to be divided between the sexes in the ratio of 105 males to 100 females, in line with recently observed trends. The number of infants aged 0 at the end of the year is calculated by taking the projected number of births, deducting the number of deaths found by applying the special infant mortality rate and adding half the number of net migrants aged 0 last birthday.
The projections are computed for each of the component countries of the UK and the results are added together to produce projections for England & Wales, Great Britain and the UK.
For the 2010-based projections, data have been published up to 100 years ahead. However, the long-term figures should be treated with great caution. Population projections become increasingly uncertain the further they are carried forward, and particularly so for smaller geographic areas.
To emphasise this uncertainty, a distinction is made between four different time periods:
For each year to 2035 (the first 25 years of the projection period), the projected population is published for both five year age groups and by single year of age.
For selected years from 2035 to 2060 (from 25 to 50 years ahead), the projected population is also published for both five year age groups and by single year of age. Further detail is available on request.
For selected years from 2060 to 2085 (from 50 to 75 years ahead), the projected population is published for five year age groups only. Further detail is available on request.
For selected years from 2085 to 2110 (from 75 to 100 years ahead) the projected population is published for five year age groups only. These results are not considered appropriate for a wide range of uses but have been made available in line with making datasets publicly available under the government's transparency agenda. Further details are available on request.
The principal population projections provide a consistent starting point for all government planning which is affected by the numbers in the population. They are based on assumptions considered to be the best that could be made at the time they are adopted. However, due to the inherent uncertainty of demographic behaviour, any set of projections will inevitably be proved wrong, to a greater or lesser extent, as a forecast of future demographic events or population structure. To give users of the projections an indication of this uncertainty, a number of variant population projections have also been produced, based on alternative assumptions of future fertility, mortality and migration.
In addition to the principal assumptions, high and low variant assumptions have been prepared for each of the three components of demographic change - fertility, life expectancy and migration. These variant assumptions are intended as plausible alternatives to the principal assumptions and not to represent upper or lower limits for future demographic behaviour.
Aside from the principal projection, nine standard variant projections are presented here. These are the six possible 'single component' variants (that is, varying only one component at a time from the principal assumptions); and two 'combination' variants: those produce the largest/ smallest total population size, and one special case scenario of zero net migration (natural change only). Eleven further variant projections will be released on 23 November 2011, see tables 1.2 and 1.3.
The 2010-based variant population projections are based on the categories of assumptions shown in Table 1-2.
|Fertility||Life expectancy||Net migration|
|Special case scenarios|
From these categories of assumptions, the variant projections released on 26 October 2011 have been created as shown in Table 1-3.
|Fertility||Life expectancy||Net migration|
|1 Principal projection||Principal||Principal||Principal|
|Standard ‘single component’ variants|
|2 High fertility||High||Principal||Principal|
|3 Low fertility||Low||Principal||Principal|
|4 High life expectancy||Principal||High||Principal|
|5 Low life expectancy||Principal||Low||Principal|
|6 High migration||Principal||Principal||High|
|7 Low migration||Principal||Principal||Low|
|Standard ‘combination’ variants|
|8 High population||High||High||High|
|9 Low population||Low||Low||Low|
|10 Young age structure2||High||Low||High|
|11 Old age structure2||Low||High||Low|
|12 High medium-term2 dependency1||High||High||Low|
|13 Low medium-term2 dependency1||Low||Low||High|
|Special case scenarios|
|14 Replacement fertility2||Replacement||Principal||Principal|
|15 Constant fertility2||Constant||Principal||Principal|
|16 No mortality improvement2||Principal||No improvement||Principal|
|17 Zero net migration (natural change only)||Principal||Principal||Zero|
|18 No change2||Constant||No improvement||Principal|
|19 Stationary2||Replacement||No improvement||Zero|
|20 No mortality improvement & zero net migration2||Principal||No improvement||Zero|
|21 Long-term balanced net migration (UK only)2||Principal||Principal||Long-term balanced|
In the principal projections, the UK projection is calculated simply as the sum of the projections for the four individual countries (England, Wales, Scotland and Northern Ireland). Therefore, the projected population numbers, deaths at each age, and births at each age of mother for the UK are just the sum of those for the four individual countries. The 'assumed' UK fertility and mortality rates are then 'back-calculated' from these projected births, deaths and population numbers.
However, it does not necessarily follow that the same process should be used for variant projections. It is intended that the standard variants for individual countries should represent broadly comparable margins of uncertainty to those for the UK. However, for migration at least, relative uncertainty tends to increase for smaller areas. In particular, for Scotland, Wales and Northern Ireland, migration flows from the rest of the UK are at least as large as international migration flows from outside the UK. These cross-border flows, therefore, account for much of the uncertainty regarding total net migration for these countries. However cross-border migration flows cannot be high in all four countries simultaneously, or low for all countries, as they must sum to zero at UK level.
For these reasons the variant migration assumptions for the UK are 'non-additive', that is, they have not been calculated as the sum of those for the four individual countries. For fertility and mortality, however, uncertainty at UK and individual country level is unlikely to differ significantly. The variant fertility assumptions and the variant mortality assumptions for the UK have been calculated directly from those for the individual countries by the same back-calculation process used for the principal projections.
Details about the current planned changes to state pension age (SPA) under the Pensions Act of 1995 and 2007 were published with the 2008-based national population projections in Changing state pension age - Tables A6-1 to A6-6. (123.5 Kb Pdf)
Any woman born before 6 April 1950 will attain SPA when she reaches the age of 60.
For women born between 6 April 1950 and 5 April 1955, the date that SPA will be attained can be found in changing state pension age table A6-1 (123.5 Kb Pdf) .
Any person born after 5 April 1955 but before 6 April 1959 will attain SPA when he or she reaches the age of 65.
For persons born between 6 April 1959 and 5 April 1960, the date that SPA is attained is shown in changing state pension age table A6-2 (123.5 Kb Pdf) .
Any person born after 5 April 1960 but before 6th April 1968 will attain SPA when he or she reaches the age of 66.
For persons born between 6 April 1968 and 5 April 1969, the date that SPA will be attained can be found in changing state pension age table A6-3 (123.5 Kb Pdf) .
Any person born after 5 April 1969 but before 6 April 1977 will reach SPA when he or she reaches the age of 67.
For persons born between 6 April 1977 and 5 April 1978, the date that SPA will be attained can be found in changing state pension age table A6-4 (123.5 Kb Pdf) .
Any person born after 5 April 1978 will attain SPA when he or she attains the age of 68.
Sources: Pensions Act 1995 Chapter 26 Part II Section 126 and Schedule 4; Pensions Act 2007 Chapter 22 Part I Section 13 and Schedule 3.
The data presented in this report does not reflect proposed further changes to the state pension age published by the government in the Pensions Bill 2011. They propose increasing the state pension age to 66 - women's state pension age will increase more quickly to 65 between April 2016 and November 2018; and from December 2018 the state pension age for both men and women will start to increase to reach 66 from October 2020. These proposed changes are not yet law and still require the approval of Parliament. Further information relating to these proposals can be found on the DirectGov website.
From 2004 to 2009, all women aged 60 to 64 are of SPA. From 2020 to 2023, all women aged 60 to 64 are of working age. From 2010 to 2019, the number of women aged 60 to 64 who are of SPA is calculated using changing state pension age table A6-5 (123.5 Kb Pdf) .
Similarly, from 2020 to 2023, all persons aged 65 and over are of SPA. From 2046 onwards, all persons aged 65 to 67 are of working age and all persons aged 68 and over are of SPA. From 2024 to 2045, the number of men and women aged 65 to 67 who are of SPA is calculated using changing state pension age table A6-6 (123.5 Kb Pdf) .
An expert academic panel advises the ONS on the appropriate assumptions to use for national population projections. This panel met in 2011 to advise on the assumptions for 2010-based population projections.
The membership of the panel is as follows:
Doctor Ann Berrington, University of Southampton,
Professor David Coleman, University of Oxford,
Professor Emily Grundy, London School of Hygiene and Tropical Medicine,
John Hollis, Greater London Authority,
Professor Carol Jagger, Newcastle University,
Professor Les Mayhew, City University,
Professor Mike Murphy, London School of Economics,
Professor Phil Rees, University of Leeds,
Professor John Salt, University College London,
Professor Robert Wright, University of Strathclyde.
A note of the panel meeting held on 17 March 2011 can be found in Appendix A.
The panel completed a questionnaire where they were asked what they thought were the most likely future levels of fertility, life expectancy and migration. They were also asked for their views on the validity and importance of a wide range of arguments which might be thought likely to influence future trends. The questionnaire was devised by the International Institute for Applied Systems Analysis (IIASA) in Vienna and was adapted by ONS for use in the UK.
Details of the panel's views on the most likely levels of the total fertility rate, life expectancy at birth and total net migration to the UK (and associated 67 per cent and 95 per cent confidence intervals) in the years 2014 and 2034, will be published in the November release.
For the previous round of projections, the panel were asked their views on the arguments for migration. In addition, the panel were also asked their views on the likely impact of the recession upon future fertility and migration trends. For this round, the panel were asked to focus on fertility assumptions and trends. They were asked to consider six forces with the potential to affect fertility levels in the long-term (to 2034). By considering a range of arguments within each force they were asked to assess the overall importance of each force and its likely impact on future family size.
'Changes in macro level conditions that influence costs of children' was rated as the most important force, and most experts felt it would have a small downward impact on fertility. This represents a change from two years previous, when this factor was expected to have little impact or a small upward impact, and illustrates the effects of the current economic climate on experts' thinking. Among the macro-level factors, housing was the issue with the greatest expert consensus on the effect on fertility, which was thought to be negative.
The forces 'Trends in ideal family size and individual desires for children' and 'Trends in patterns of education and work', although rated as second and third in importance, were on balance felt to have little impact on the overall level of future fertility. This was often due to arguments having opposite effects that balance each other out. For example under the education/work force, the positive impact of increasing flexibility of work may be outweighed by the negative impacts on fertility of reduced grandparent availability for childcare due to changing female employment at older ages. Similarly, experts did not expect average ideal family size to decline significantly, despite childlessness becoming increasingly common (as a valid lifestyle choice or a constraint due to lack of a stable partnership), as they agreed that the perceived value of siblings was strong enough to maintain the desire for many families to have at least two children.
'Population composition', rated fourth, was expected to have a small upward influence on fertility overall, mainly due to the higher fertility of some migrant groups which was expected to persist. However the downward impact on fertility of care for elderly parents also came across clearly in the responses.
The new principal projections are based on the long-term assumptions of future fertility, mortality and net migration (that is, immigrants less emigrants) and are summarised in Table 1-4. These have been agreed in consultation with NRS, NISRA and the Welsh Government. Comparisons are given with the assumptions made for the previous 2008-based projections.
|United Kingdom||England||Wales||Scotland||Northern Ireland|
|Fertility – Long-term average number of children per woman|
|Mortality - Expectation of life at birth in 2035 *|
|Net migration† – Annual long-term assumption|
For the UK, the long-term assumption of average family size is 1.84 children per woman, the same as that for the 2008-based projections. The long-term UK fertility assumption in official national projections has been lowered on a number of occasions since the 1960s baby boom, but it was increased for the 2006-based projections, reflecting the overall increasing fertility rates seen in all four countries of the UK over the past decade. The assumption has been kept at 1.84 since then, despite the Total Fertility Rate (TFR) continuing to rise, because an increasing TFR may result from changes in the timing of childbearing in different cohorts of women and does not necessarily mean that completed family size will increase.
The 2010-based projections see a change from the 2008-based projections in terms of improvements in mortality. It has been assumed for the principal projection that over the first 25 years of the projection period (by 2035), annual rates of mortality improvement will converge to a common rate of 1.2 per cent a year for those born in 1939 and after, with those born in the years 1925 to 1938 assumed to experience higher rates of improvement. Conversely, for those born before 1924, annual rates of improvement are assumed to be lower than 1.2 per cent in 2035. Overall, these are the same assumptions for the rates of mortality improvement in the target year as those used in the 2008-based projections (where the target year was 2033) for those born before 1940; for those born in 1940 and later the proposed improvement rates in the target year of 1.2 per cent are higher than the 1.0 per cent assumed in the 2008-based projections.
Projected period life expectancies at birth for the year 2035 are around 0.1 years lower than in the previous projections for males and 0.1 to 0.3 years lower for females. These differences are mainly due to the age-specific mortality rates for 2010 being assumed to be higher and the rates of mortality improvement between 2010 and 2011 assumed to be lower at many ages below 90 compared to those projected for the same period in the 2008-based projections. Over the early years of the projections these counterbalance the assumption of higher rates of mortality improvement at most ages in 2035.
The new long-term assumption for net migration to the UK is +200,000 each year compared with +180,000 a year in the 2008-based projections. This increase is due to taking account of final migration data for two new years (2008 and 2009).
Details of the policy governing the release of new data are available by visiting www.statisticsauthority.gov.uk/assessment/code-of-practice/index.html or from the Media Relations Office email: firstname.lastname@example.org
These National Statistics are produced to high professional standards and released according to the arrangements approved by the UK Statistics Authority.
Notes of meeting held on 17 March 2011
Location: Office for National Statistics (ONS), Myddleton Street, London
John Hollis, Greater London Authority
Suzie Dunsmith, Population projections
Adrian Gallop, Population projections
Ben Humberstone, Deputy Director ONSCD
Julie Jefferies, Fertility and Family Analysis Unit
Julie Mills, Mortality and Ageing Unit
Esta Clark, General Register Office for Scotland (GROS)
Keira Murray, GROS
Louise O'Leary, Fertility and Family Analysis Unit
Steve Rowan, Population projections
Victoria Staples, Population projections
Denise Williams, Population projections
Professor Emily Grundy, London School of Hygiene & Tropical Medicine
Professor Mike Murphy, London School of Economics
1.1. Ben Humberstone welcomed everyone on behalf of ONS and gave a brief introduction noting that the panel had three new members. Suzie Dunsmith outlined the timetable for the assumption setting and consultation process for the 2010-based population projections.
1.2. This was the fourth meeting of the national population projections expert advisory panel, which is now considered to be an integral part of the national population projections assumption setting process. The main aim of this meeting was to have an informed discussion about the long-term assumptions for the forthcoming UK-level population projections.
1.3. The role of the expert panel is strictly advisory. It was emphasised that ONS was not seeking a consensus where none exists and wanted to hear the full range of views on the key assumptions. Though all opinions would be considered, the final decision on the assumptions to be adopted for the next set of projections rests with ONS and the Registrars General.
1.4. The 2010-based population projections are scheduled for publication in October 2011.
1.5. ONS would give introductory presentations on each of the three sets of assumptions - fertility, mortality and migration, after which the floor would be opened up for general discussion. A questionnaire on the assumptions had been sent to all of the experts and some high level summary results were circulated before the meeting. The questionnaire results would be published later with the note of this meeting.
2.1. Julie Jefferies' fertility presentation began by looking at recent trends in fertility. She illustrated how the UK Total Fertility Rate (TFR) had been rising steadily since 2001, reaching 1.96 in 2008, but fell slightly in 2009 to 1.94. This was slightly higher than in the 2008-based principal projection, but moving in the direction predicted. A similar trend was observed in all four UK countries. The principal 2008-based projection was based on a scenario where the recession would have a small downward impact on fertility in the short-term with greater falls in fertility among women in their twenties than older women. Projected fertility remained fairly high with a completed family size in the long-term of 1.84. She noted that entirely trend-based projections are not realistic for fertility, so some judgement is needed on plausible trends for different age and parity groups.
2.2. She then went on to look at data for 2010. Although data for the whole year is not yet available, quarterly data shows that the number of births have started to increase again from the fourth quarter of 2009. However, there were differences between countries with births in Scotland falling slightly.
2.3. With reference to age-specific fertility rates in the UK, it was shown that there have been continued increases since the mid-1970s for women in their thirties and forties. However the rise in fertility for women aged 30-34 stopped between 2008 and 2009. Rates have been declining steadily since 1980 for women in their twenties. However, since 2001 there has been a recovery in rates for women aged 25-29, except in 2009, and a bottoming out of the falling rates for 20-24 year olds. Analysis of results from the questionnaire sent to the expert panel indicated that in the short term the panel expected these trends to continue, with fertility decreasing or stabilising for women under 30 and increasing or stabilising for women aged over 30.
2.4. Julie then compared the TFR as it's normally calculated from age-specific rates with the TFR calculated from estimated parity progression ratios, calculated using a life table method. Since 2005, the TFR calculated by parity has been lower than the published TFR suggesting that the published TFR could be slightly over-estimating the level of fertility. She observed that progression ratios show that there has been a notable increase in progression to first birth since 2001. So it seems likely that part of the increase in the TFR since 2001 is due to timing of first births. It was noted that this was not likely to be sustainable, particularly if the pool of childless women diminishes. She asked the panel if they had a view on this.
2.5. In terms of cohort fertility, with reference to achieved family size, evidence for women born from 1940 to 1964 in England and Wales shows that completed family size has been declining gradually in each successive cohort. However, the 1970 cohort has almost converged with the 1965 cohort by age 39 and the 1975 cohort with the 1970 cohort by age 34. Recent cohorts have in general experienced higher teenage fertility but lower fertility in their early twenties than the 1965 cohort, but they are catching up in their late twenties and thirties. Therefore it appears that decreases in completed family size seem to have slowed/stalled over the last 5 years or so.
2.6. Julie reported on preliminary analysis of the expert panel members' responses to the questionnaire. The average of the experts' estimate of the TFR in 2034 is 1.79, with 8 respondents ranging between 1.8 and 1.9, suggesting that the long-term assumption used in the 2008-based projections of 1.84 is still considered to be the right level by the majority of the panel. However, there was less agreement between respondents about the likely level of the TFR in 2014. Five respondents thought the TFR in 2014 would stay around the current level (1.90-2.0), four a little lower (1.80-1.90) and one somewhat lower (1.65-1.80). With an average of 1.87, this is similar to the value used for 2014 in the 2008-based projections of 1.85.
2.7. She briefly summarised some of the panel's responses about forces that act on fertility levels and the impact and likelihood of these forces:
The factor the panel judged to be most important on average was changes in macro level conditions that influence costs of children. Most felt that this would have a small downward impact on fertility. This is a change from 2 years ago, when trends in education and work was felt to be the most important factor which shows the impact of current economic climate on our thinking. Among the macro-level factors, housing was the issue where there was greatest consensus about its the effect on fertility, which was thought to be negative.
Ideal family size and ‘education and work’, although important, were on balance felt to have little impact on the overall level of future fertility.
Population composition was expected to have a small upward influence on fertility overall, mainly due to migration. However the downward impact on fertility of care for the elderly also came across clearly in the responses.
The force with the least consensus was the effect of partnership change and family networks where equal proportions thought this factor would have an upward or downward impact on future fertility.
2.8. The panel discussed the possible effect of the recession on fertility and whether it was possible to learn from looking at previous recessions. It was noted that the current recession was not necessarily over, as it is not a simple spike and that it was still visible in current unemployment rates, and, on past experience, its affects could last for a further 4 years. The impact of previous recessions on fertility is hard to ascertain and if fertility were to continue to fall it could be due to the current recession or alternatively, because tempo catch-up is running out.
It was suggested that previous recessions were different, since this time there has been a large fall in GDP but the rise in unemployment has not been as large. There is one theory that in a recession birth rates may go up as female employment decreases. However there was no evidence that this happened in the last recession as more men lost jobs than women. Evidence from research on the 1970 cohort of women showed that they postponed fertility more than perhaps they would have done if their economic position had been more secure. Some experts felt that this may be different this time, as public sector cuts in the next few years are expected to result in more women losing their jobs. There was some agreement that in the short term, if a women was out of work this may encourage them to have more children especially if approaching the end of their fertility. Also that the effect of the recession is likely to vary depending on age and parity.
A recession adds to uncertainty, generally leading people to behave more conservatively in their spending and saving. One expert thought that the rapid rise in fertility has come to a stop, and that it was safe to assume that fertility is not likely to rise, especially with reductions in government spending.
2.9. Discussing recent short term trends, it was noted that the fall in fertility in 2009 was not large and there has been some evidence of recovery in 2010. One expert thought that fertility levels have now caught up with intentions for women in their thirties. In the last year there has been a flattening of the upward trend for rates of women in their 30s; this could be due to the catch up petering out or the recession, but it would be difficult to say precisely.
2.10. It was suggested that it would be useful to look at the composition effect of UK and non-UK born mothers on fertility, looking at trends in age specific fertility rates (ASFR) for these groups to see if this helps explain changes in fertility levels observed. One expert questioned what impact births to Polish women in the UK have had on fertility rates since 2004. It was noted that fertility in Poland is very low but that they may behave differently in the UK compared with fertility behaviours at home. Another expert noted that this may have something to do with salary differentials between countries.
Whether there is there a second generation effect on UK-born mothers born to non-UK born parents was discussed. It was noted that a lesson learnt from developing ethnic projections was that country of birth should have been included in analysis to better understand trends.
However it is expected that fertility of ethnic minorities born in the UK would converge to UK norms. It was noted that some ethnic groups continue to have different fertility levels, for example some work has been carried looking at the fertility of Hasidic Jews who despite low incomes tend to have very large families.
2.11. One expert observed that the recession has had a different impact on fertility in London than elsewhere, where fertility rates fell more. They have observed increased fertility in older women and those born outside the UK. He noted that A8 Citizens will be eligible for benefits and asked whether this would have any effect. (From end April 2011, accession states will have been EU members for 7 years and become entitled to same rights as other EU citizens.)
2.12. The impact of availability of affordable housing was discussed. Some research done a while ago, using the longitudinal study shows that there is higher fertility in social housing. It was thought that middle income women may think more about affordability in their 30s and that this could delay fertility for this group. For women born in the 1970s and 80s we would expect some recuperation at older ages, but it was questioned as to whether the disadvantaged will be able to recuperate.
2.13. The group were questioned about the possible impact of spending cuts on fertility. Firstly what the panel thought would be the effect on fertility rates of under 20s if funding is withdrawn from projects and programs previously tasked to reduce levels of teenage pregnancy. Also cuts to the Sure Start and other support programs for families and changes to tax credits may all affect fertility in the future. It was noted that the availability of affordable childcare may become more important and that the impact on fertility should be considered.
2.14. The panel discussed the impact of affordability of child care on fertility. Research carried out in Sweden and Norway produced conflicting conclusions. A paper using Norwegian data that show a linear relationship between availability of free child care and fertility, showing a relationship between more childcare and more babies. However, research in Sweden found that there was a greater relationship between employment & income and increased fertility than with availability of childcare. It was noted that in London the TFR increased in times of increased availability of affordable childcare, but there were also other factors to consider which could have caused increases, for example more Asian & African women in London.
2.15. The group then discussed what they thought future trends in fertility rates in the short and long term would be. There was some agreement that the rise in fertility in the last 10 years is unlikely to be sustained. One expert thought that a TFR of 1.84, the long-term assumption used for the current projections, is not plausible in the long term, irrespective of the economy, and that the TFR is likely to be closer to the number of children women say they want (intentions), that is 2.00. Another thought that the least plausible trend in the short term is a sharp decline, noting they had heard of greatly increased maternity bookings in South Hampshire for births in October 2010, when the fourth quarter births are published this will say whether this is true or not, they also asked if ONS have any data on expected deliveries.
2.16. A factor noted as likely to restrict further increases in fertility to those over 35 is the increased risks to congenital malformations at older ages, however it was noted that this may result in increases in fertility in early thirties.
2.17. The group discussed whether changes in partnership status of women could explain changes in trends. It was noted that evidence from 1958 cohort and 1970 cohort studies shows that the factor most significant in whether a women did recuperate is whether or not they were in a partnership (relationship), after controlling for other factors. It was noted that there was no spike in marriages at the turn of the century although there was a peak around 2003/04.
2.18. Julie noted that when setting the assumptions last time, that when the suggested ASFRs were projected forward to the long term they did not match with the suggested long term TFR trend. To make them consistent meant either increasing rates at older ages or reducing rates for younger ages. She sought the group's view on what was likely to happen to age specific rates.
2.19. One expert thought that it was important to understand what has happened since 2001 to cause rates to increase and that if we can understand the reasons behind the change between the 1990s and 2000s, then we may be able better reason when it will change again. It was noted that the increases to fertility rates have also been observed in other countries over the same time.
3.1. Adrian Gallop's mortality presentation began by illustrating the trend of increasing period life expectancy from 1850 to the present day. In particular, it was noted that period life expectancy(LE) at birth improved especially rapidly during the first half of the 20th century, since when the rate of increase has slowed. In contrast, period expectation of life at age 65 has exhibited a fairly steady linear increase for females. Male expectation of life at 65 lagged someway behind females during the 1950s to 1970s, but then saw steeper increases thereafter so it now appears to be catching up to where it would have been had it in risen in line with increases in females LE at age 65, due possibly to lifestyle changes.
3.2. The UK mortality projections involve estimating mortality rates and trend rates of mortality improvement by age and gender for the base year of the projections, setting rates of mortality improvement for a specific year in the future (the 'target year') and then making assumptions on the method and speed of convergence from the current improvement rates to the target rates and thereafter. A potential new method for smoothing historical improvement rates to produce estimates of current rates of improvement using p-splines was discussed. Current average rates of improvement are about 2.8% for men and 2.0% for women. Historical and future age-specific mortality rate improvement was illustrated with the use of lexis diagrams which showed the annual improvement in age-specific mortality rates since the 1960s. Peak levels of mortality improvement appear to be a special feature of the generations born between 1925 and 1945 (centred on the generation born around 1931). It was explained that the current mortality projections methodology involves projecting improvement rates by cohort for older generations, but by age for younger generations where there is no evidence of cohort effects taking place. Comparing the latest years improvements with 2008-based projections indicates we were possibly a little too optimistic in the initial improvement rates used, which would feed through to future years.
3.3. The target rate of improvement used in the 2008-based projections is 1.0% pa for most ages (rising to 2.5% for those born around 1931). The 1% figure is based mainly on being the average annualised rate of improvement over the 20th century and also on the fact that, using these rates, the improvements over future periods from the base year 2008 were similar to, or greater than, those experienced for the same number of years going back from the base year. There is continuing debate whether future technical, medical and environmental changes will have a greater or lesser impact than in the past.
3.4. There are a large number of potential drivers for future mortality change such as changes in bio-medical technology, effectiveness of health care systems, behavioural changes related to health (decline in smoking prevalence, obesity), emergence of new diseases and re-emergence of old diseases , environmental change, changes in population composition and so on. The questionnaire responses rated the first three of these as being the most important factors. The decline in mortality from circulatory diseases over the last 40 year was noted and that for women the standardised mortality rate for cancer was now higher than that for circulatory diseases. Going forward it may be that it is what happens to cancer mortality rates that will have the largest influence on future improvements over the longer term.
3.5. Life expectancy trends were examined within an international context. It was shown that amongst developed countries UK life expectancy currently ranks around average with UK female life expectancy at birth ranking lower than UK male life expectancy at birth. However, it was noted that, for those countries compared, by 2050 only the projected period life expectancy at birth for males in Switzerland and the high life expectancy projection for Australia were higher than that projected for UK males in 2050. For females, there are more countries with a projected higher LE in 2050 than projected for the UK. However, given the corresponding positions in 2008, the overall amounts of projected mortality improvement to 2050 for males and females was somewhat higher than that projected in the official projections for many other developed countries. It was also shown that within many developed countries, male to female life expectancy differentials had been falling in recent years.
3.6. Responses to the expert questionnaire were presented: expectation of life at birth in 2034 was on average estimated at 83.8 years for males which would have been obtained with a target improvement rate of 1.25% in 2033 in the 2008-based projections and 87.2 for females (1.14% target improvement) by the panel experts. These compare with the current projection for 2034 of 83.3 years for males and 87.0 years for females. However, there was considerable variation between respondents with three estimates being rather higher and one rather lower, being outside the 67 per cent confidence interval for the average of the group.
3.7. It was observed that men born in the 1950s appear in general to have experienced lower rates for mortality improvement or even worsening mortality than those of earlier generations. Adrian noted that the higher male mortality observed in the late 1980s to 1990s for these cohorts then aged in their 20s-30s, was in part due to HIV AIDS. The panel asked if it was possible to quantify how much of the improvement in male mortality rates is due to changes in occupations, for example a reduction in numbers of people employed in hazardous industries. The philosophy behind using past trends going back a long time was questioned as older observations come from a different history in terms of economic, medical and industrial conditions also causes of death that are no longer relevant. Adrian responded that it should be remembered that the projections are not forecasts. Part of the justification for the choice of target rates arises from looking at what would happen if trends go forward in the same way as they as go back. He noted that the panel’s average target improvement rates were around 1.33 for males and 1.18% for females. These were a little more optimistic than last time for males and broadly similar for females (last time the panel was not asked to provide separate improvement rates for males and females), but still more optimistic than the 1% improvement used for past projections. Adrian noted that the assumptions were not created by projecting mortality by cause of death; but that cause of death data were looked at to help explain what had happened in past years and what might happen going forward. For example, whilst mortality from cardiovascular disease was a major contributor to declining mortality rates, this would be less so in future as mortality rates from heart disease etc fell below those from cancer.
3.8. One expert noted that if the initial rates are based on an unusual year, the current method can give false figures into the future. Adrian said there can be edge effects in the current methodology where rates of improvement for the first year of the projections are derived from data up to the year preceding the projection base year and extrapolated forward one year, even after smoothing. It was observed that we were probably a little too optimistic in 2008. The new methodology for smoothing historical data had been tested to see the effects adding in a further year's data had on the improvement rates derived for the most recent years used. It was found that stepping back a couple of years from the latest year gave improvements which changed by lower amounts when data for a new year were added. Adrian noted that if this was followed there would have to be some method to decide the improvement rates from that year to the base year of the projections. Another expert agreed that this is a sensible approach particularly at the end of an intercensal period.
3.9. The panel discussed reasons for the past improvements in mortality rates. Generalising, some experts felt that the decline in deaths due to circulatory diseases is probably mainly due to changes in lifestyle although there were other factors as well. Decline in cancer deaths would be due more to improvements in medical treatment and technological developments than lifestyle changes (with the exception of smoking). It was asked if it is possible to break down change (reduction in mortality) to particular medical advances for example cancer screening, new medications for specific types of cancer. A panel member said that recent research shows that there have only been small improvements for males aged under 50 years old with more rapid improvements for those aged over 50, particularly those aged between 71 and 82 (2.5% improvement). It was observed that the gap between male and female expectation of life at age 65 (EOL65) is closing, and it was asked if they will eventually be the equal or even cross over. Would improvements continue at the same rate? If we look at other countries such as Japan, is there a biological factor that would provide limits on improvements to mortality? Adrian said that extrapolating current trends would result in the EOL65 for men and women converging, then crossing over. He also noted that if different target improvement rates are used for men and women and that for males were higher than that adopted for females this would cause a methodological problem in that the differential in life expectancy would reverse with males having a higher life expectancy than females. Another panel member noted that the variability in the past appears to have disappeared therefore they were fairly confident that the trend would continue.
3.10. The panel generally agreed that obesity was more likely to have an impact on morbidity than mortality rates. It was noted that in the USA obesity was routinely medically monitored and subsequent interventions would possibly change outcomes. It was agreed that at younger ages obesity would not affect mortality. However, it is not yet understood what would be the impact if people stay obese for a long period of time for those aged 50-70. One expert noted some research that showed that losing years of life depends on what age the person becomes obese and for how long they remain obese. The Foresight report looking to 2040 suggested obesity reduced mortality by 12-18 months over the period to 2040 but this was against a general increase in life expectancy due to other factors of 6-7 years over the same period. It was suggested that a review of existing research is needed on whether obesity makes you ill or causes premature death.
3.11. Adrian sought the panel's view about how to model future mortality of the very old (those over 100) and whether the currently projected large improvements in later years as those born around the early 1930s reached these older ages were reasonable - these lead to quite high reductions in mortality rates at the oldest ages over the projection period. The Gompertz curve doesn’t fit the mortality data at the oldest ages well. Research suggests that at the oldest ages mortality rates tend to flatten out at around 0.5-0.6 rather than continuing to increase with age. It was noted that it was difficult to model what was happening at the oldest ages or project this forward because data are sparse. One expert thought that there may be a ceiling on life expectancy, another commented that mortality in this age group has little impact and that focus should be on those aged 80-100 years.
4.1. Suzie Dunsmith's migration presentation began by examining trends of in-, out-, and net migration focusing on recent years. Net migration to the UK has followed an upward trend since 1991, with peaks in 2004 (245,000) and 2007 (233,000) corresponding to the expansion of the EU. Latest estimates show net migration fell to 163,000 in 2008, rising again to 198,000 in 2009 with this rise due to a decrease in emigration compared with the previous year. Net migration in 2009 was similar to levels seen in 2005 and 2006, which is in line with the average over the last five years of 200,000. This compares to the long-term net migration assumption of 180,000 in the 2008-based projections, with a high variant of 240,000 and low variant of 120,000.
4.2. In 2009 UK net international migration was 198,000 of which 44,000 was attributed to net out migration of British and 242,000 to net in migration of non-British citizens. There were 567,000 immigrants in 2009, with 83 per cent (471,000) non-British, and two thirds (303,000) from non-EU countries. Emigration from the UK was 368,000, with emigration of British nationals at its lowest since 1999 (140,000). The increase in net migration in 2009 was driven by an increase in non-British in-migration and low levels of emigration.
4.3. The reasons for migration were considered, with work a key reason given and an increase in in-migration for study. Trends in National Insurance Number (NINo) allocations to overseas nationals show an increase in NINo allocations to migrants from EU accession countries. IPS flows of A8 and A2 citizens show decline in inflows and increasing outflows in the most recent years, and if this is a continuing trend there is the possibility of over-estimating long-term net migration.
4.4. Suzie reported preliminary findings from responses to the expert questionnaire. The panel estimated in-migration in 2034 to be on average 537,000, compared to 579,000 in the last projection round. Emigration in 2034 was estimated to be similar to the 2008 figure, on average 381,000 compared to 379,000. Net migration in the short-term to 2014 was estimated on average as 171,000. The panel estimated net migration in 2034 to be 162,000, which compared to the average of 200,000 over the last five years, suggesting the panel thinks that net migration will fall from current trends in the long-term.
4.5. Suzie then went through the questions for discussion by the panel. These included the impact of the global economy upon net migration; the impact of the points-based system on net migration; and how the likely future of the European Union will affect UK net migration.
4.6. The discussion began with the questioning of modelling of future population projections on net migration flows. Previous projections methods used to treat migration inflows and outflows separately, and academic literature does not use net flows. In research developing ethnic population projections models, a panel member had explored using emigration rates that aligned initial rates to current rates and assuming immigration flows. Emigration rates could be used since the population at risk was known. It was suggested that as using this method changed the resulting projections that this methodological uncertainty should be investigated. The panel noted that using net migration in the projections methods was discussed at the previous two expert panel meetings, and required further investigation
4.7. It was suggested that the potential impact of an ageing population on migration should be considered. The dependency ratio fell in 2007 after a long period of stability, and there is a need for employment and workers for caring roles needed to sustain an older population. It was suggested that this underlying dynamic would drive in-migration, implying an increase in net migration, but the dependency on current government policy was noted. Year-to-year variability was highlighted, and it is important to consider when projecting to 2034.
4.8. A panel member commented that they had found the migration component of the questionnaire the hardest section to answer as international migration is hard to predict. Germany was given as an example of a country where migration has both risen and fallen. The panel noted that migration was highly dependent on political change, and a panel member suggested that there was not a workforce need for international migration, referencing work by Philip Martin. Changes to rules affecting students and caps on visas were given as other examples of policy change and pressure. In the past academic institutions encouraged overseas students for financial reasons, but the new caps could cause financial difficulties for these institutions, which in turn may increase pressure to change policy further.
4.9. The Government now has an explicit policy to reduce in-migration, which has not been the case for previous projection rounds. It was noted that migration is a function of geographical boundaries, and with no control over the EU as a policy entity it was suggested that the focus should be on migrants from countries outside Europe.
4.10. A panel member noted that it was harder to reduce net migration than increase it. The panel were hesitant to put a figure on future migration and stressed the importance of different scenarios. A panel member noted that the panel had agreed in their questionnaire responses assuming net migration of over 100,000 until 2014. This suggested that the panel do not think that the government will meet their target of reducing net migration. The potential sensitivity of using these figures in projections should be considered.
4.11. The discussion moved to the Points-based system (PBS) and its effect on migration. It was noted that the PBS does not apply to migration from the EU. It was suggested that the future expansion of the EU, for example to include the Ukraine, would be in the time frame of this round of projections and had the potential to increase the number of migrants considerably. A panel member suggested that particular industry sectors, for example universities, the care industry and restaurateurs would put pressure on the Home Office to make special cases in the PBS for immigration. The expert thought that business pressures would shift the view of migration. It was noted that the entrance of the primary economic migrant could be controlled, but there was no way to control the flow of dependents. Canada was given as an example where points in their system were lowered if they wanted to increase the population.
4.12. A panel member suggested that events in other countries should be considered, for example after 9/11 there was an increase in the number of students coming to the UK. The potential effect of restrictions on EU accession migrants ending in April 2011 was discussed, with a panel member suggesting that migrants may move to countries closer to home. It was suggested that other factors that could affect migration, such as environmental change, should be considered.
4.13. One panel member was suggested more attention should be placed on the zero net migration projections in presenting the projections by comparing how the projections change as the value of the net-migration is varied. It was suggested that baseline net migration of 180,000 was too definite, but was something that the media could focus on at publication. A panel member disagreed and suggested that baseline net migration should be higher. Another panel member thought that in the short-term there would be no change in net migration.
4.14. The importance of migration to Scottish population growth was highlighted. A panel member noted that migration was critical, but mortality had little real effect, as there were more deaths than births in Scotland and migration was supporting population growth. The importance of migration in Scotland has economic consequences, as if employers can't recruit workers they will leave, but this was thought to be a few years away in other parts of the UK.
4.15. A panel member asked if ONS were going to implement stochastic projections to tackle the issue of uncertainty, and why ONS research into these methods had stopped. The panel discussed these methods, highlighting the probability ranges produced as an indication of uncertainty, and the potential difficulties in meeting user needs (for a single figure) using this method. Work by Wolfgang Lutz (Vienna Institute of Demography) suggested that the median value was usually indicated which would be equivalent to the principal projection. It was suggested that these methods would remove the need to make assumptions about the future net migration figure. A panel member commented that stochastic forecasts showed greater variability in projections of younger populations than older populations. Progress from early stages of ONS research had been published as a research report. Other government departments had not requested stochastic forecasts, but did use variant projections. The 2010-based national population projections will use current methods, with a clear explanation of why stochastic methods were not used. ONS will consider the issue of stochastic forecasting outside this meeting.
4.16. The panel discussed the effect of the global economy. Recent OECD research has seen an increase in international migration. The effect in the short-term has been a slight slow-down of in-migration, and lower emigration. The global population is projected to increase by 2.5 billion by the middle of the century, and using the UN migrant definitions, between 2.5 and 3 per cent of the population are defined as migrants. The UK share of this future increase needs to be considered. A panel member suggested that the exchange rate could also affect migration. An example was A8 migrants coming to the UK when the exchange rate was good and returning home with the devaluation of the pound. A panel member suggested that geographic breakdowns should be changed to reflect whether the countries are rich, poor, or developing.
4.17. Ben Humberstone asked the panel to consider emigration. It was suggested that emigration depends on the type of migrant, suggesting that changes to the exchange rate would likely only have an effect on short-term migrants. The issue of Britons abroad was raised, a panel member suggesting that there may be retirees not emigrating at the moment, but this may increase when the economic situation improved. The large number of British citizens living in Spain could have great impact on services if they return to the UK.
4.18. Recent efforts by ONS in measuring short-term migrants were noted. A panel member commented that this had implications for the population being projected; if long-term and short-term migrants were being included. There is a perception that the population includes all people, and there is a need to consider whether alternative populations including short-term migration and visitors could be useful.
4.19. The need for estimates of flows by individual country of origin was highlighted, but it was noted that the IPS is currently not able to deliver flows between the UK and all other countries. Migration questions on e-borders were discussed. A panel member noted that data on emigrants from Britain was collected in other countries and there is a need for this information to be collated.
Population projections unit