1. Introduction

Population estimates for England and Wales

The Office for National Statistics (ONS) produces annual estimates of the resident population of England and Wales as at 30 June every year. The estimates are provided broken down by local authority, sex and age. The most authoritative population estimates come from the census, which takes place every 10 years in the UK. Population estimates from a census are updated every year until the next census to produce mid-year population estimates.

Population estimates for the UK

ONS produces population estimates for England and Wales. Estimates from Scotland and Northern Ireland are also collated by ONS to produce UK totals. Estimates for Scotland are produced by the National Records of Scotland (NRS), while the Northern Ireland Statistics and Research Agency (NISRA) produces the estimates for Northern Ireland. Estimates for each of the UK constituent countries are compiled using a common methodological approach and aim to be as consistent as possible. Details of the specific data sources and methods used across the UK are summarised in a UK Comparisons note.

This document relates to the estimates for England and Wales only.

A guide to the methodology used to produce the mid-year population estimates for Scotland is available from the NRS website. Details on the methodology used to create the Northern Ireland population estimates are available from the NISRA website.

Usually resident population

Population estimates refer to the usually resident population. This can mean that estimates of population do not necessarily coincide with the number of people to be found in an area at a particular time of the day or year.

For most people, defining where they usually live for the purposes of the census, for example, is quite straightforward. For a minority of people the concept of usual residence is, however, more difficult to define, for example, for students, members of the armed forces, prisoners and international migrants.

Specific rules are used for these groups:

  • higher education students and schoolchildren studying away from home are resident at their term-time address

  • members of the armed forces are usually resident at the address where they spend most of their time

  • prisoners are usually resident in the prison estate if they have a sentence of 6 months or more

International migrants are usually resident if they intend to stay in England and Wales for more than 12 months.

Protection against disclosure

The estimates are produced using a variety of data sources and statistical models, including some statistical disclosure control methods, and small estimates should not be taken to refer to particular individuals.

Quality assurance

The quality of the mid-year population estimates are consistently monitored by ONS. This includes quality assurance of the administrative and survey data sources that are used to calculate the estimates; the statistical methods applied to produce the estimates and the tables of data published on the ONS website. Further details are available in the Annual mid-year population estimates QMI (Quality and Methodology Information) and in the Quality Assurance of Administrative Data (QAAD) reports, listed in Annex 1.

Back to table of contents

2. Cohort component method

Rolled-forward estimates

Population estimates are produced using a cohort component method. This is a standard demographic method that uses high quality data sources to inform components of population change. The three major components of population change are summarised as follows.

  1. Natural change (births, deaths and ageing)

    The starting point for producing the estimates is the count of resident population from 30 June of the previous year. This population is aged on by 1 year. Births during the 12-month period are added to the population, while deaths during the period are removed according to recorded age, sex, and usual area of residence.

  2. Migration

    Movement of people into and out of the UK (international migration) and movements between different areas in the UK (internal migration) are also accounted for in the population estimates. Note that internal migration includes both cross-border moves between the other countries of the UK and moves between local areas within each part of the UK. Migration is the most difficult part of the estimate process to measure precisely because the UK has no population register. Rather, we use the best proxy data available on a nationally consistent basis to estimate migration.

  3. Special populations

    Adjustments to the population estimates are made for some special population groups that are not captured by the usual internal or international migration estimates: members of the armed forces and prisoners. These populations have specific age structures, which remain fairly constant over time. They are not aged-on with the rest of the population. Such populations are referred to as static populations.

Census-basing

The method in this section describes how mid-year population estimates are calculated for years when there is no census. For years in which there is a census, the mid-year population estimates are based on the census estimates and therefore, a slightly different approach is necessary. Rather than ageing on the population by 1 year, the population is only aged on by the period of time between the census and 30 June. Similarly, the components only need to account for change during this period rather than a whole year.

Research and development

We continue to research ways of ensuring and improving the quality of the population estimates, including analysis of new data sources that become available. The latest information on research into new and improved methods for producing our standard population statistics is available.

Uncertainty estimates have been created to give users additional information of the quality of these estimates. Measures of statistical uncertainty are available for the years mid-2012 to mid-2015.

In-depth methodology

The following sections describe in more detail how we estimate the components of population change in the mid-year estimates produced for England and Wales.

Back to table of contents

3. Births

Change in population due to births

Births in England and Wales occurring between 1 July of the previous year and 30 June of the current year are added to the population at age 0, by sex and allocated to the local authority of usual residence of the mother.

Births data

Data on live births by sex are obtained from the Civil Registration System administered by the Office for National Statistics (ONS) and are based on births occurring (and then registered) in England and Wales. As registration of births may legally take place up to 42 days after a birth, the data received refer to the date of birth rather than the date of registration.

Births to mothers outside England and Wales

The Civil Registration System captures information on all births in England and Wales. This includes births to mothers who are usually resident elsewhere, but not necessarily those births to mothers who are usually resident in England and Wales that take place elsewhere.

We assume that the number of births for the two groups are similar in number and on average balance each other out. In this way, births to non-usually resident mothers are added to the population estimates as a proxy for those births elsewhere to usually resident mothers. We impute local authorities of residence for these births using the distribution of births we know about during the year.

Back to table of contents

4. Deaths

Change in population due to deaths

Deaths that are registered in England and Wales between 1 July of the previous year and 30 June of the current year are subtracted from the population by sex, age and local authority of usual residence.

Deaths data

Deaths data are obtained from the Civil Registration System administered by the Office for National Statistics (ONS). The data are supplied by sex, age and local authority of usual residence in England and Wales. To be consistent with the mid-year reference date we adjust age at death to 30 June.

The Civil Registration System captures information on all deaths in England and Wales. This includes deaths to people usually resident elsewhere (outside England and Wales). In the calculation of subnational population estimates these people are allocated to a local authority, imputed using the distribution of deaths by age and sex we know about during the year.

The Civil Registration System does not record deaths of usual residents of England and Wales that have occurred abroad and which are not registered in England and Wales. These deaths are excluded from the deaths data and do not feature in the calculation of the mid-year population estimate.

Unknown local authority of residence

Local authority of residence is not recorded for a number of deaths. For these, a local authority is imputed using the distribution of deaths by age and sex we know about during the year.

Late registrations

We make a small adjustment for anticipated late registrations to allow for deaths that were not registered at the time the data were extracted. The number of late registrations in the previous year is used as a proxy for late registrations in the current year given the assumption that the number of late registrations does not vary much year-to-year.

Back to table of contents

5. Internal migration

To account for migration of people within the UK, data are obtained for flows of migrants between each pair of local authorities in England and Wales as well as flows of migrants between England and Wales and the rest of the UK (so called cross-border flows).

Internal migration data

Internal migration estimates are primarily based on data that flags up when people change their doctor as they change address. Since most people re-register with a new doctor after moving, these data are considered to provide a good proxy indicator of migration. Similar data sources are used both for cross-border flows and moves within England and Wales.

A combination of three administrative data sources are used in this way as a proxy for internal migration within England and Wales: the National Health Service Central Register (NHSCR), the Patient Register Data Service (PRDS) and Higher Education Statistics Agency (HESA) data.

NHSCR data

The NHSCR records the movements of patients between former health authority areas (HAs) and is combined with PRDS data held by individual former HAs to produce estimates of migration between local authorities.

Similar data sources are used to obtain estimates of cross-border flows to and from Scotland and Northern Ireland. The total flows to and from constituent countries of the UK are agreed between the Office for National Statistics (ONS), the National Records of Scotland (NRS) and the Northern Ireland Statistics and Research Agency (NISRA), based on records of in-migration to the relevant country.

PRDS data

Each former HA holds lists of patients registered with GPs. ONS gets a snapshot of data extracted from each area’s Patient Register as at 31 July each year. This reference date is based on the assumption that it takes about a month to register with a GP and hence appear on the Patient Register after moving to a new area. This enables migration estimates to be produced for the year ending 30 June.

By obtaining an extract from each Patient Register on an annual basis and combining all the extracts together, a total Patient Register for the whole of England and Wales is created. Duplicate and temporary NHS records are removed from the register when combining the extracts and a small number of missing data fields are imputed in order to improve data quality.

The records are compared between the current year and the previous year, and this enables the identification of people who have changed their postcode during the period. For the purpose of estimating the population, it is assumed that a person who changes their local authority of residence between one year and the next is a migrant.

Reconciling NHSCR and PRDS

Unlike the NHSCR, the PRDS provides migration estimates down to local authority level. However, the PRDS has one major limitation: it cannot capture the migration of those who move during the year who were not registered with a GP at one of the two mid-year time points.

The largest group missed in this situation are migrant babies aged less than 1 year. This is because they don’t appear on the GP record at the previous mid-year. Similarly, international in-migrants won’t have been registered with a GP at the previous mid-year and so be missed if they moved between registering with a GP and the current mid-year. Other missing data arises in cases where people move during the year but do not appear on the PRDS at the current mid-year point, for example, because they have emigrated. This is also the case for deaths.

To address these issues, data are combined with more complete information in the NHSCR and then used to produce the migration estimates for local authorities. Since migrant babies aged 0 are not captured by the Patient Register data, these moves are estimated by combining the data for 0 year olds in the NHSCR with the local authority distribution of moves of 1 year olds in the PRDS.

The NHSCR data source was discontinued in February 2016. As a consequence England and Wales internal migration estimates for 2016 have been calculated by combining the 2016 PRDS data with the 2015 NHSCR data.

HESA data

One of the known limitations of relying on GP registration changes is that young people, particularly young men, can be slow to change their registration when they move. Given one of the most common reasons for migration among young people is to attend a course at a higher education establishment we use HESA data to supplement GP data. HESA data contains records of students at higher education establishments in England and Wales. Data are available for both domicile address and term-time address, which allows potential internal migrants to be identified and an adjustment to be made to the internal migration estimates.

Internal migration and special populations

Movements of members of the armed forces are not included in the internal migration estimates. Whilst the NHSCR records movements of people into and out of the armed forces, movements of serving members are not recorded. For similar reasons, movements of prisoners are also not included in the internal migration estimates. The population of armed forces and prisoners are estimated separately.

Changes to methods

The method for calculating internal migration estimates will undergo further planned changes before the 2017 mid-year estimates are produced. Details of these changes can be found in Appendix 2.

Back to table of contents

6. International migration

An international migrant is defined as a person who changes his or her country of usual residence for a period of at least a year. International migration estimates are made up of immigration, emigration, asylum seekers and refugees plus their dependants. Estimates of international migration exclude the armed forces, whose movements are estimated separately.

International Passenger Survey

The Office for National Statistics (ONS) national estimates of international migration are based on the International Passenger Survey (IPS). The IPS is a long-running ONS survey that operates at UK ports of arrival and departure. The IPS is the only source of data on UK migration that is specifically designed to identify people who change their country of usual residence for at least 12 months. This is consistent with the usual residence definition for international migrants in the population estimates.

Sample survey

The IPS is a sample survey and so only a sample of migrants to or from the UK are interviewed. Within this sample, only a small proportion will be long-term international migrants.

The IPS is based on voluntary, face-to-face interviews with a sample of passengers travelling via airports, sea routes and the Channel Tunnel. The migrant respondents sampled are scaled to produce national migration estimates using a complex weighting system. Because the IPS is a sample survey, the results are subject to a degree of statistical uncertainty.

Limitations of use at local authority level

The IPS estimate is not reliable enough to be used at local authority level, as the sample of people in each local authority is too small – with many local authorities having little or no IPS respondents.

Furthermore, a migrant’s initial intentions about where they will settle may not be realised. For example, there is a tendency for in-migrants to state an intention to migrate to London, but actually settle in another part of the UK. Therefore, the IPS estimate at local authority level may not be a reliable indicator as to where people migrate from or to.

Local authority estimates of international migration are derived using a variety of data sources and methods as described in the following methodology for immigration and emigration.

Visitor switchers and migrant switchers

The IPS does not take into account the changing intentions of passengers. Some migrants intend to remain in or out of the UK for 12 months, but actually go on to spend less than a year. These are called migrant switchers. Other migrants intend to remain in or out of the UK for less than a year but actually spend longer. These are called visitor switchers.

Migrant and visitor switchers are identified by the IPS as they complete their journey. The passenger is asked how long they intended to stay in the UK or overseas when they initially arrived or departed, and for how long they actually remained in or out of the UK. An estimate is calculated for the proportion of migrants or visitors who changed their intentions on the duration of their stay.

The likelihood of a visitor changing their intentions can vary depending on their citizenship and place of last or next residence. Therefore, visitor switchers are split into four groups: those entering the UK who are European Economic Area (EEA) and non-EEA citizens, those leaving the UK who are EEA citizens going to the EU, and all other citizens leaving the UK going to anywhere in the world. However, unlike visitor switchers, there is no distinction between citizenships or countries of last or next residence for migrant switcher calculations.

International migration estimates are adjusted for migrant and visitor switchers, as part of the methodology for immigration and emigration.

Immigration

Local authority level estimates

Streams

The IPS total data for England and Wales is streamed, mainly by reason for migration (for example, worker, student, other) and relevant administrative sources are used to distribute immigrants to each local authority. Record linkage is used both within and between the administrative sources to minimise definitional differences and duplication.

Data sources:
  • Migrant Worker Scan (MWS): provides a count of foreign nationals applying for a National Insurance Number (NINo); this is the main source used to distribute immigrant workers

  • Higher Education Statistics Agency (HESA) data: used for distributing publicly funded higher education students and private higher education students

  • administrative data sources from Department for Business, Energy and Industrial Strategy (BEIS) and Welsh Government (WG): used to distribute further education student immigrants from the EU

  • Home Office visa data: numbers of non-EU international migrants at educational institutions with a Certificate of Acceptance to Study and data on non-EU international migrants granted Leave to Remain status: used to estimate the distribution at local authority level of non-EU international migrants who are further education students

  • census data: used for distributing UK-born returning migrant flows

  • Patient Register Data Service (PRDS): data on migrants who register for a GP whose previous address was outside of England and Wales: used alongside the other administrative sources to distribute the remaining immigrants, such as children, those aged 17 to 59 who are not students or workers and those aged 60 and over

Sex and age breakdown

Census cluster analysis

2011 Census data on immigrants has been used to group local authorities into clusters with similar age and sex structures. Census immigrants are defined as those who stated that they had arrived in England and Wales in the year before the census and intended to stay more than 12 months. Recent research, carried out by ONS, has confirmed that the 2011 Census is still the best available data source for sex and age distributions of immigrants at local authority level.

Sex and age breakdown

Each year, the cluster analysis of immigrants is used to inform the sex and age distribution of the local authority immigration estimates. Each local authority level record is assigned to the cluster group based on local authority, and sex and age is imputed based on the average sex and age distribution of that cluster group.

Emigration

Local authority level estimates

Modelling the estimates

A statistical model is used to estimate emigration at local authority level. The model estimates the numbers of emigrants over the year using relationships established between the estimate of emigration from the IPS and estimates from other data sources (covariates) that can predict a more robust estimate of emigration at local authority level than the IPS alone can provide.

Response variable

The IPS weighted estimate of emigration is used as the response variable and is based on a 3-year average; the year estimated and the previous 2 years. This is necessary because the number of emigrants sampled in 1 year is small and the spread of the sample across the country is uneven with many local authorities having no sampled emigrants.

Covariates

The covariates used in the model come from census, administrative and survey data sources and have been found to have a strong relationship with emigration at local authority level. Each covariate is fixed in the model to avoid problems of instability in the year-on-year emigration estimates.

Additionally regional New Migration Geography for Out-migration (NMGo) indicators that account for geographical design issues in the survey data not already picked up by the model are included.

Constraining

The modelled estimates of emigration at the local authority level are constrained to the IPS estimates at NMGo level. IPS estimates at NMGo level are first averaged over 3 years (the year estimated and the previous 2 years) to ensure that the estimate is robust, and then constrained to the region totals for the year of estimation.

Sex and age breakdown

A combination of 2011 Census and IPS data are used to add sex and age detail to the modelled local authority emigration estimates. This process groups local authorities into clusters based on sex, age and citizenship data gained from the census.

Local authority cluster analysis

Census data for immigrants are used to classify local authorities into clusters that have similar patterns of UK born and non-UK born migration, as a proportion of the total population.

We use census data to classify local authorities into groups as IPS data is not reliable at local authority level. Immigration data is used because census data on emigration is unavailable to produce estimates of emigration. We assume that immigration data is likely to show similar patterns of UK born and non-UK born migration to that exhibited in emigration patterns at local authority level.

Citizenship and sex

Using the local authority clusters derived from census data, IPS data for emigrants are used to create a distribution by citizenship (British, non-British) and sex, for each cluster. The IPS is broken down by citizenship because it is assumed that British and non-British emigrants are likely to have a different age structure.

Single year of age

Three years of IPS data (current year and previous 2 years) are used to provide a detailed single year of age distribution, by citizenship and sex. This age distribution is smoothed using a centred average to remove noise.

The smoothed single year of age distribution is applied to the cluster, citizenship and sex distribution.

Applying to modelled emigration estimates

The local authority level emigration estimates are assigned a cluster group. Citizenship, sex and age are then imputed onto each record, based on cluster group.

Changes to methods

The method for calculating international emigration estimates will undergo further planned changes before the 2017 mid-year estimates are produced. Details of these changes can be found in Appendix 2.

Asylum seekers and dependants

Most movements of asylum seekers are not captured by the IPS. The UK Border Agency of the Home Office provides ONS with data on asylum seeker applications and their dependants, including removals, refusals, withdrawals and appeals. This information is used to adjust the estimated international migration inflows and outflows for asylum seekers.

Any asylum seekers counted by the IPS on arrival or departure to or from England and Wales are excluded from our processing and Home Office data for asylum seekers used to ensure flows are not double-counted.

Asylum seeker inflows

Estimates of asylum seeker flows into England and Wales are based on the number of asylum seeker applicants. An adjustment is made for the small number of asylum seekers who are recorded as both a principal applicant and a dependant. Counts of applicants who returned to their country of origin within a year of their application are removed.

Asylum seeker outflows

Data on asylum seeker flows out of England and Wales are estimated based on asylum applicants that are assumed to have left England and Wales after staying for at least a year, using information on removals, refusals and withdrawals.

Local authority

Estimates of asylum seeker inflows and outflows are only available by region. Data for asylum seekers at local authority level are available for those who receive support from the National Asylum Support Service (NASS). There are no data sources that provide local authority level information for those who claim asylum but do not request any associated support. Estimates of asylum seekers are broken down from region to local authority using a broad assumption that 60% receive support and 40% are unsupported. The 60% assumed to receive support are distributed to local authority level using the local authority distribution for asylum seekers receiving support for accommodation. The remaining 40% are assumed to have the same geographical distribution of residence as those given subsistence-only support by NASS (for example, those who are not dispersed to accommodation in particular areas).

Single year of age

A single year of age distribution for both asylum seekers and their dependants is derived using a combination of data from the Home Office and the national mid-year population estimates from the previous year.

Refugees

The Home Office operates a number of international resettlement schemes that result in people entering the UK. Such people are granted humanitarian protection, and while they are commonly referred to as “refugees”, it is important to note that they do not have refugee status according to the strict UN definition. The Home Office database covering such schemes includes the Gateway Protection Programme, the Mandate Scheme and Syrian Vulnerable Persons Resettlement Scheme (SVPR), but not those resettled to the UK under the “ex gratia scheme” for Afghan locally engaged civilians.

In September 2015, the government pledged to receive 20,000 Syrian refugees over the subsequent 5 years. Consequently flows of refugees became sufficiently large to justify refugees being treated as a separate component within the calculation of the mid-year population estimates for England and Wales.

Refugee inflows

For the purpose of the mid-year population estimates, refugee inflows are estimated by single year of age and sex for each local authority in England and Wales.

Most movements of refugees into the UK are not captured by the IPS; any that are counted are excluded from processing to ensure that no double-counting occurs. ONS is sent an extract of the resettlement database by the Home Office for people arriving in the year up to 30 June of the reference year. The data includes information on the age, sex and citizenship of the refugees.

For those refugees entering under SVPR, information on their initially assigned local authority is also available. These refugees are allocated to a local authority in advance of resettlement and this has been recorded on the Home Office dataset that ONS receive. Analysis on the actual regional distribution of SVPR persons within England and Wales supports the use of the initially assigned local authority as their place of usual residence.

For a minority of refugees, information on their initial local authority of residence is not known. For these, a local authority is imputed using the distribution of SVPR refugees within England and Wales.

Refugee outflows

Outward flows of refugees are thought to be very small at this stage; refugees leaving the UK would be covered by the IPS, as they would leave from a UK port of departure.

Back to table of contents

7. Home armed forces

Who’s included?

The population estimates include all members of the UK armed forces (UKAF) who are stationed in England and Wales. Members of UKAF deployed on operations and temporary assignments overseas are also included in the population estimates where their last permanent station is in England and Wales. Personnel that are serving on overseas postings are removed from the population estimates, but we account for their flows and those of their accompanying dependants into and out of England and Wales.

Special population

UKAF are treated as a special population as the movements of military personnel are not captured by the data sources used to estimate international and internal migration.

It is assumed that UKAF personnel and their dependants travel by military flights into and out of England and Wales when serving in posts overseas; routes that are not covered by the International Passenger Survey (IPS). It is also assumed that UKAF personnel are not on GP registers and are therefore not counted in the internal migration estimates. However, it is assumed that dependants are on GP registers, so movements of dependants within England and Wales are not part of the special population.

Home armed forces data

The Office for National Statistics (ONS) receives aggregated UKAF data from the Ministry of Defence (MOD). This data includes military personnel counts by age, sex and local authority of base.

ONS also receives aggregate data from British Forces Germany (BFG) by sex and age, of dependants (partners and children) who accompany members of UKAF stationed in Germany. Germany has the second largest population of UKAF after the UK, and accounts for approximately three-quarters of all UKAF posted overseas.

Census data for the home armed forces are also used to inform distributions for local authority of usual residence.

Change in UKAF stationed in England and Wales

Data are obtained from MOD for UKAF by sex, age and local authority of base, stationed in England and Wales. To fit in with the population estimates usual residence definition, the UKAF population is estimated at the residence at which they spend most of their time. A base to residence distribution based on census data is used to adjust UKAF from their local authority of base to their local authority of residence.

Change in civilian population

Any change in the population of UKAF from one year to the next will be reflected in civilian population of England and Wales – those joining and leaving the UKAF will create a resulting inflow and outflow between UKAF and the civilian population.

This flow between the UKAF population and the civilian population must also take account of UKAF serving overseas if they are usually resident in England and Wales.

A reduction factor is applied to all UKAF (including those stationed overseas) to estimate those who would be usually resident in England and Wales, as opposed to other parts of the UK. The proportion of UKAF (excluding those stationed overseas) stationed in England and Wales is used as a proxy for calculating this reduction factor.

To account for the change in the population of UKAF stationed in England and Wales, the previous year’s estimated population is subtracted from the current year’s estimated population, by sex, age and local authority of usual residence.

A local authority of residence is imputed for each net flow using a local authority distribution derived from the census for the permanent home of members of the home armed forces.

Change in overseas dependants

It is assumed that dependants (partners and children) of members of the UKAF who are serving overseas are not picked up by the IPS and are therefore treated as part of the home armed forces special population.

BFG data on dependants accompanying UKAF stationed in Germany are used to provide a ratio (number of dependants per UKAF member) and sex and age distribution that can be applied to UKAF serving overseas to estimate the overseas dependant population.

A reduction factor is also applied to the estimated overseas dependant population to estimate those who are usually resident in England and Wales. The reduction factor is calculated using the same proportion as previous, so that only the overseas dependants who are usually resident in England and Wales are estimated.

In order to account for the change in the overseas dependant population, the current year’s estimated overseas dependants population who are usually resident in England and Wales is subtracted from the previous year’s overseas dependant population, by sex and age.

A local authority of residence is imputed for each net flow using a local authority distribution derived from the census for members of the home armed forces living with a partner.

Compilation

In order to calculate the total change of home armed forces, we calculate by sex, age and local authority:

Net change in UKAF stationed in England and Wales plus net change in the civilian population plus net change in overseas dependants

Back to table of contents

8. Foreign armed forces

Special population

Foreign armed forces based in England and Wales are treated as a special population in the population estimates as it is assumed that they are also not picked up by the methods used to estimate internal and international migration.

It is again assumed that foreign armed forces personnel travel by military flights into and out of England and Wales; routes that are not covered by the IPS. It is also assumed that foreign armed forces personnel are not on GP registers and are therefore not counted in the internal migration estimates.

Who’s included?

All foreign armed forces usually resident in England and Wales should be included in the population estimates. The United States Air Force (USAF) make up the majority of foreign armed forces; however, there are a number of military personnel from other US service arms (US Army, Navy and Marine Corps) that are also based in England and Wales.

The foreign armed forces component only accounts for military personnel from the USAF, with the exception of a small adjustment made for other US service arms currently located in Harrogate and North Kesteven. Foreign armed forces that are not from the US are not accounted for as part of the special population as there are no data currently available. However, these are considered very small in number.

United States Air Force (USAF) data

Data for USAF based in England and Wales are supplied to the Office for National Statistics (ONS) annually on or around the reference date of 30 June for the number of USAF military personnel, by sex, age and base in England and Wales.

Adjustment for Harrogate and North Kesteven

An adjustment is made for the local authorities of Harrogate and North Kesteven for other US service arms to account for pockets of localised foreign forces resident in these local authorities. The adjustment is based on data from the US Department of Defence’s Statistical Information Analysis Division (SIAD) on the total number of US Army, Navy and Marine Corps personnel based in the UK.

Base to local authority of residence

The population of England and Wales is estimated at the local authority of usual residence. USAF data is only provided by base and therefore local authority of usual residence is imputed using data derived from the 2011 Census. For any bases in the USAF data where there is no base to residence information available in the census, residence is assumed to be at the local authority of the base. This is a valid assumption as the majority of members of the US armed forces live on base.

Change in foreign armed forces population

The change in the foreign armed forces population between the two mid-year points is estimated by subtracting the previous year’s estimated foreign armed forces population from the current year’s estimated foreign armed forces population, by local authority of residence, sex and age.

Assumptions not stated elsewhere

Further assumptions are made in how we estimate the foreign armed forces special population. It’s assumed:

  • joiners and leavers of the foreign armed forces population are not taken from or put back into the general England and Wales population

  • dependants of members of the foreign armed forces are picked up by the usual methods used to estimate international and internal migration and are not treated as part of the special population

Changes to methods

The method for calculating estimates of foreign armed forces dependants will undergo further planned changes before the 2017 mid-year estimates are produced. Details of these changes can be found in Appendix 2.

Back to table of contents

9. Prisoners

Special population

Population estimates include all prisoners imprisoned in England and Wales with a sentence of 6 months or more. Prisoners are treated as a special population in the population estimates as it is assumed that movements of people into and out of prisons are not picked up by GP registers used to estimate internal migration.

Prisoners data

The Ministry of Justice supplies data on the number of people resident in prisons in England and Wales on 30 June of the reference year, by prison location, sex and age. For the purposes of the population estimates, a person is regarded as usually resident in a prison if they have been sentenced to serve 6 months or more.

Change in prisoners population

Change in the prisoner population between the two mid-year points is estimated by subtracting the previous year’s estimated prisoner population from the current year’s estimated prisoner population, by local authority, sex and age. This change can only be indicative as the prison estate population can fluctuate widely between mid-year points due to operational needs.

Change in non-prisoner population

Any change in the estimated prisoner population from one year to the next will be reflected in the general population of England and Wales – those joining and leaving the prisoner population will create a resulting inflow and outflow between the general population.

In order to distribute inflows and outflows of prisoners to and from the general population of England and Wales, the local authority distribution of the previous year’s population estimate is used and we distribute flows to the local authorities with the highest populations.

Foreign national offenders and offenders from other parts of the UK

The prisoners component of the population estimates assumes that all prisoners in England and Wales remain in England and Wales following the completion of their sentence. Foreign national offenders who are deported following completion of their sentence or ex-prisoners who move to other parts of the UK are not accounted for in this method. Owing to difficulties in accurately estimating this population, it is assumed that the flow of ex-prisoners returning to England and Wales from elsewhere balances out these flows.

Back to table of contents

10. Compilation

National (England and Wales) and subnational (local authority) estimates from each component are compiled to produce subnational and national population estimates.

The previous year’s population estimate by sex, age and local authority of usual residence is aged on by 1 year. The number of births between the two mid-year points is added into the population at age 0. Deaths between the two mid-year points are removed from the population estimates.

Net flows of international and internal within UK migration are then added into the population estimates. Changes because of special populations are also added into the population.

The resulting population estimate is the final population estimate for 30 June of the current year, by sex, age and local authority.

Flow diagram of the Cohort Component Method

Previous year’s population estimate
   ↓
Age on by 1 year
   ↓
Add births between 1 July and 30 June to age 0
   ↓
Subtract deaths between 1 July and 30 June
   ↓
Add net internal and international migration between 1 July and 30 June
   ↓
Add changes due to special populations (summarised together in the estimates)
   ↓
Current year’s population estimate

Special populations

Arrows 1 and 2 show the process used to calculate the change in each special population between the previous year and current year. The main concept is that people move from the general population into a special population (for example, when joining the armed forces) and when they leave a special population they return to the general population. Arrows 3 and 4 show the process used to calculate the change in the England and Wales population that is due to a change in the special population between the previous year and current year.

Back to table of contents

11. Appendix 1: Quality assurance of administrative data reports

We quality assure the administrative data used in the estimation of the annual mid-year population estimate to ensure that they are suitable for this purpose. To gain further insight on data quality issues and the impact on population statistics, please see the quality assurance of administrative data documents for each data source, which are available on the ONS website:

Births

Deaths

UK Armed Forces

US Armed Forces

Patient Register (PR)

Higher Education Statistics Agency (HESA)

Prisoners

National Health Service Central Register (NHSCR)

Migrant Workers Scan (MWS)

Asylum Seeker Data and Non Asylum Enforced Removals

Home Office Immigration

Asylum Seekers Support

Back to table of contents

12. Appendix 2: Proposed changes in methods

This appendix contains descriptions of a number of proposed changes in the methods used to produce the population estimates for local authorities in England and Wales. It is divided into three sections:

  1. Internal migration

  2. Emigration

  3. Dependants of foreign armed forces

We propose to implement these changes in the estimates for mid-2017, due to be published in June 2018, and to publish at that time a back-series of revised estimates from 2011 to 2016 incorporating, as far as possible, these new methods.

We welcome any comments or questions (email to pop.info@ons.gov.uk) on these proposals.

  1. Internal migration

    The largest element of population change at the local authority level is internal migration – that is, people moving from one local authority in the UK to another. The methods for estimating internal migration were significantly improved in estimates for 2012 published in 2013 when the availability of new Higher Education Statistics Agency (HESA) data allowed a simpler and more reliable approach to be adopted for estimating migration of students. We are proposing two further developments of the methods.

    Improved models for the destination of students after they leave higher education

    The current methodology is thought to be an improvement on the previous method in that it is much more accurate in estimating migration of students to their place of study. However, there is still scope for improving methods for estimating the destination of students who move after leaving higher education. On leaving higher education and not updating their Patient Register record (PR) they will either stay in the local authority in which they lived while studying, or return to their address as recorded on the PR (their pre-study home address, or PR address), with an increasing probability of having moved to their PR address as time goes on.

    The proposed methodology uses patterns of internal migration observed in past higher education leavers to impute moves to current higher education leavers who do not update their PR (we are defining higher education leavers as those who are on the HESA dataset one year, but not on it the following year).

    The new methodology addresses two of the limitations of the current system. Firstly, the current system only gives two possible destinations for higher education leavers who do not update their PR records. They can either be left at their study local authority, or moved to the local authority of their PR address (on the assumption that this is their “home local authority” where they lived before moving to study). This misses all the higher education leavers who have come from one local authority, studied in a second local authority and, on leaving higher education, gone to live in a third local authority (these moves are only identified when the PR record is updated). Secondly, students moving to a higher education establishment are increasingly encouraged to register with a local GP or medical centre and as such, their PR address is likely to reflect their study local authority rather than their home local authority. Thus when we return them to the local authority of their PR address on the assumption that this is their “home local authority, we may in fact be leaving them at their study local authority.

    Our research suggests that most higher education leavers update their PR within 3 years of their move. Therefore, to establish the pattern of moves that higher education leavers make, we will take the first change of address (strictly speaking, output area) on their PR over a 3-year period as their new location. If there is no change of address over the 3 years, we will assume that they have remained in the same area (accepting that some people will have moved but not updated their PR even over this longer period). Analysis of these changes of address allows us to create a matrix of destinations for each local authority by sex (female higher education leavers may have different internal migration patterns to males). So, for example, for female higher education leavers living in Portsmouth when they left higher education, we can determine that a% went to Southampton, b% went to Cornwall, c% stayed in Portsmouthand so on.

    We can then establish a pool of current year higher education leavers. For those that have updated their PR, we have evidence of their post-higher education address and thus will remove them from the higher education leavers’ pool to be adjusted and treat them the same as the rest of the population. For those that have not updated their PR, we will move them to a new location based on the migration matrices developed previously described. So, using the previous example, for those people leaving Portsmouth who have not updated their PR, we would move a% of them to Southampton, b% to Cornwall, leave c% in Portsmouth and so on. Those that have been moved will be identified on the stock files as a higher education leaver, so that we don’t move them again until they update their PR.

    The impact of this change of methods will be:

    • more accurate population estimates, and age profile, of people in their twenties for areas that attract a large number of graduates
    • a corresponding improvement in the estimates for areas that were the pre-study home local authority for students
    • the creation of some spurious moves as imputed destinations are updated following updated Patient Register records (these spurious moves will slightly inflate the internal migration estimates but will not have a material effect on the population estimates as they “net out”) (similar spurious moves, but with different geographical patterns, also exist under the current method)

    Improving estimation of migration moves within the year not identified by comparing addresses at the start and the end of the year

    As described previously, the current methodology uses NHS Central Register (NHSCR) data to adjust the initial estimates to allow for moves of people who were only present at either the start or the end of the year, but not both.

    The NHSCR data source is now no longer maintained and this provides both the requirement to change, and the opportunity to improve, the current method of estimating this type of move.

    A weakness with the NHSCR data is that it was only available at the Family Health Service Authority (FHSA) level, which is at a less detailed scale than local authority. For instance, the North West Lancashire FHSA contains Blackpool, Fylde, Preston and Wyre, so the same scaling factor will be applied to all four, although each may have different internal migration patterns. This is likely to be more of an issue where FHSAs contain local authorities with significantly different characteristics (for example, urban or rural, or universityor non-university).

    In addition to the geographical limitations, the current scaling method does not take account of age or sex, so the scaling applied to, say, 50 year old males is the same as that applied to 21 year old females. However, the two are likely to have different internal migration behaviours.

    We are proposing replacing the existing “scaling” method of estimating this type of move with an alternative approach of directly identifying (a subset of) these moves by linking the start-year and end-year population stocks file with registrations of births and deaths.

    So taking the example of births. When a baby is born, it will be added to the mid-year estimate (MYE) population at the location of its mother’s normal residence given on the birth certificate. However, it may move location between its time of birth and the time the mid-year PR stock file is taken. The baby cannot be on the PR stock file before it was born, and so there is no reference point from which to detect a move. To illustrate this, let’s take a sequence of events:

    1. 2015 PR stock extract is taken on 31 July 2015

    2. Baby born on 3 September 2015 to a mother living in Portsmouth

    3. Baby moves to Cornwall on 3 April 2016

    4. Baby registered with doctor in Cornwall on 20 April 2016

    5. 2016 PR stock extract taken on 31 July 2016

    When we are doing the 2016 MYE we will add one 0 year old to the population of Portsmouth.

    We will not see an internal migration move for the baby as, although they are in the right location of Cornwall on the 2016 PR, they are not on the 2015 PR so we will not see a move.

    However, we can link the 2016 PR to the births recorded in the last 12 months and where the location on the PR differs from the location on the birth record, we can record a move. So in this example, we would create a move from Portsmouth to Cornwall, thus reducing the population of 0 year olds in Portsmouth by one and increasing the population of 0 year olds in Cornwall by one.

    A similar logic can be applied to people who move in the MYE period before they die, but linking the 2015 PR with the death records. An example would be:

    1. A 50 year old person is living in Durham

    2. 2015 PR stock extract is taken on 31 July 2015

    3. Person moves to Newcastle on 20 September 2015

    4. Person registers with doctor in Newcastle on 13 December 2015

    5. Person dies on 20 February 2016

    6. 2016 PR stock extract taken on 31 July 2016

    In this case, using the existing method, we would remove a 50 year old from Newcastle when we do the 2016 MYEs. However, this person was in Durham at the last MYE. We will not see an internal migration move as the person is not on the 2016 PR as they have died. However, if we compare the 2015 PR with the death records, we can record a move from Durham to Newcastle that “corrects” the population estimates.

    It is recognised that similar logic could be used for people who immigrate to England and Wales and then move within the same year and, conversely, people who move within England and Wales and then emigrate within the same year. However, immigrants and emigrants are not individually identified from a data source in the same way as we have with births and deaths so we cannot link these to the relevant PR stock files to identify these moves. We have therefore assumed that the net effect of any such migrant moves on an area’s population estimate is zero. The practical effect of any such moves is likely to be negligible compared with other sources of uncertainty in the estimates.

    This approach also means that “multiple moves” within a year – for example, someone moving from Hackney to Southampton to Portsmouth to Cardiff during a year – are not estimated. This sequence of moves would be recorded as a single move from Hackney to Cardiff as it is this change that makes a difference to the population estimate.

    The impact of this change in methods will be:

    • a change in the target concept of the internal migration estimate, from that of the total number of moves in or out of an local authority to the adjustments required to mid-year population estimates to reflect moves
    • substantially fewer estimated moves, as multiple moves within a year are not included in the estimates
    • more accurate estimates of the net impact of internal migration on the population of an area

    The change in target concept implies a discontinuity in the time series of internal migration estimates (but not the population estimates) between 2010 to 2011 and 2011 to 2012. Please contact us (pop.info@ons.gov.uk) if you need further information to support your use of these figures.

    Quality-assuring the Personal Demographic Service (PDS) data source as an appropriate replacement for the Patient Register (PR) data source when the latter is discontinued

    The PR data source currently used in producing the internal migration estimates is due to be closed in autumn 2018. However, the alternative PDS data source now available to us seems to have several advantages over the PR. We have started quality-assuring the PDS data and checking that it is appropriate for use in producing the estimates with a view to moving to using the PDS data in the mid-2017 population estimates. We will publish a quality report on the use of PDS data in population estimates in due course.

  2. Emigration

    While the International Passenger Survey (IPS) is used to estimate emigration at the national and regional levels, the number of migrants sampled is not sufficient to derive reliable direct estimates of emigration at the local authority level. Since 2010, these estimates have been produced by using a poisson regression model to distribute the IPS-based national emigration estimate.

    This model is based on the relationship between weighted IPS estimates of emigration at the local authority level and a number of covariates that are associated (statistically) with a “risk” of emigration.

    In 2014, we started investigating whether improvements on that model were possible. There were two strands to this work:

    • considering improvements to the current modelling approach and whether changes in the structure of the model or the choice of covariates could produce an improved model
    • investigating whether a similar approach to that used to distribute the immigration estimates was possible; this would allocate different streams of emigration directly to local authorities using administrative sources

    We concluded that we did not have suitable administrative data to adopt the second approach and therefore looked to improve the current modelling approach. This work has resulted in three proposed changes.

    Modelling rates rather than counts

    The current model uses covariates expressed as counts to model emigration as a count. The proposed model includes an “offset term” representing the (previous year’s) population of an area. This transforms the model from a model of counts to a model of rates – in effect modelling the “risk” of a resident emigrating over the year. This is a standard approach adopted for such models and ensures that the modelled emigration remains related to the population at risk.

    Removal of New Migration Geography for Out-migration (NMGos)

    The current model incorporates an element of constraining to the IPS results for New Migration Geographies (NMGos). These are non-standard aggregations of local authorities that were originally developed to reflect the tendency of migrants recorded in the IPS to report their origin or destination in the UK as a major city rather than a neighbouring local authority but are used in the current model to make as much use as possible of detailed IPS data.

    This element of constraining can, however, lead to some unexpected results, where emigration in an area is estimated to fall, say, because of changes in covariates for a neighbouring area (that lead to an increased estimate for that area and thus require a fall in the first area’s estimate to remain consistent with the NMGo total).

    We have therefore decided to remove the NMGos from the model, with constraining done only at the regional level.

    Changing the covariates

    The selection of the covariates (that is, the explanatory variables used to model emigration) is done by a combination of stepwise selection (a statistical algorithm for selecting variables that contribute most to a model) and some manual selection of additional variables.

    The covariates in the proposed model are listed in the following table.

  3. Dependants of foreign armed forces

    We have acknowledged some issues with the population estimate for Forest Heath (and, to a lesser extent, some other areas). These issues stem from the US Air Force (USAF) population in the area. The USAF personnel themselves are represented in the estimates through their inclusion as a special population. Their dependant family members are, conceptually, covered through the conventional methods of estimating components of population change.

    In practice however, these dependants – amounting to some 9,000 across England and Wales – have relatively little interaction with the standard administrative sources used in the population estimates process and, as a result, the usual methods of estimating migration do not work well for this population subgroup.

    The practical effect of this is that the dependants present at the time of the 2011 Census are generally aged on within the area, rather than being replaced by new arrivals. This leads to anomalously low counts of females in the 20s to early 30s bracket from uncounted new dependants and anomalously high counts of females in the late 30s to 40s, and children, due to those dependants counted in the last (2011) census being aged onward through the population count without being migrated out.

    Though the size of this population subgroup is small relative to the total population in England, its geographical concentration in Forest Heath means that this methodological weakness does affect the population estimates, and particularly the age profile of the population, for that area.

    The proposed approach is to extend the scope of the existing “foreign armed forces special population” adjustment to include dependants. This adjustment will rely on USAF data already available to us.

    The impact of this change of method will be:

    • more accurate population estimates – particularly for the number of children and the age profile of women – for Forest Heath and other local authorities containing or neighbouring USAF bases
Back to table of contents

Contact details for this Methodology

Neil Park
pop.info@ons.gsi.gov.uk
Telephone: +44 (0)1329 444661