Producing UK armed forces veteran statistics for England and Wales, research update: 2023

• This research uses evolving methods to provide new data for 2023 on the population of England and Wales who have previously served in the UK armed forces.

• By linking veteran data from the Service Leavers Database (SLD) and Census 2021, we identified that 3.3% of the population aged 16 years and over in England, and 4.1% in Wales, had previously served in the UK armed forces, as of 30 June 2023.

• Like the results of the previous feasibility research into this method, these proportions are lower than the Census 2021 estimates for England and Wales (which were 3.8% and 4.5%, respectively).

• The differences between the Census 2021 figures and our current approach reflect differences in methodology and linkage, as well as deaths in the larger cohorts of the older veteran population between Census 2021 and mid-2023, relative to the total population.

• Local authority patterns in the linked dataset for 2023 align with Census 2021 estimates, with the highest proportions of veterans found in Gosport (12.1%), North Kesteven (10.3%), Fareham (8.9%), Plymouth (7.9%), and Wiltshire (7.3%), reflecting areas with a strong military presence.

• The methods used to produce the 2023 linked dataset are still being developed; as linkage rates vary across population subgroups, further research is required to better understand the impact of this on the data.

Method Development

The Office for National Statistics (ONS) has been undertaking research to provide regular statistics about the veteran population of England and Wales. The only accredited official statistics currently available are from Census 2021 (see our UK armed forces veterans, England and Wales: Census 2021 bulletin). People who have previously served in the regular or reserve UK armed forces are often known as veterans. In this article, we refer to usual residents aged 16 years and over who have served in the UK armed forces as the veteran population.

In our Feasibility research on producing UK armed forces veteran statistics for England and Wales: 2021 article, published in June 2023, we set out the user need for ongoing veteran statistics that our research aims to address, as well as future developments we wanted to explore. We have consulted with the Ministry of Defence (MoD) regarding the development of this methodology.

This article updates this research, presenting veteran population figures for 2023, which may be of use to some users where timeliness is a priority. However, further research is required to understand the accuracy of the data, particularly when broken down by age and sex, and by local authority. As well as more recent data, the method now makes use of the Demographic Index (DI) for linkage. Quality limitations and areas for further development are discussed in the article. More information about future developments and how to provide feedback, is available in Section 8: Future Developments.

Scotland and Northern Ireland have devolved responsibility for producing veteran statistics, so are not covered by this research. However, we continue to engage proactively with colleagues in the devolved governments who are also researching this topic.

Linkage Methodology

We used two data sources to identify the veteran population and aim to meet the Harmonised standard for previous UK armed forces service, Census 2021 and the MoD's Service Leavers Database (SLD). Census 2021 included a question asking people aged 16 years and over if they had previously served in the regular or reserve UK armed forces, or both. Imputed Census records were treated as having "missing" Census information. The SLD provided information on service personnel that have left the UK armed forces since 1975 and have not rejoined, as of 30 June 2023. This information was cleaned, deduplicated and deidentified before use in analysis.

Census 2021 has provided us with a valuable source of data on veterans as of Census Day (21 March 2021), while the SLD covers veterans who left the UK armed forces at any time since 1975. The expected benefit from using both data sources to produce veteran statistics going forward is that Census 2021 data can be rolled forward. Important data from the Census included older UK armed forces veterans who left service before 1975, and are therefore not covered in the SLD. In turn, the SLD included data on those who left the armed forces in more recent years, particularly as we move further away from Census 2021.

Figure 1 provides a visual representation of the linkage process. We used our 2023 Statistical Population Dataset version 5.1 (SPD V5.1) for England and Wales, filtered to those aged 16 years and over, as our population base for linkage, which allowed us to identify those veterans remaining in the population in 2023. The SPD is a dataset produced by the ONS that aims to approximate the usually resident population using a combination of linked administrative data sources. This has a reference date of 30 June 2023. It contains around 44 million usual residents of England and 2.4 million usual residents of Wales. These totals are lower than our official population estimates for those aged 16 years and over in mid-2023, which were 47.2 million and 2.6 million, respectively. More information is available in Section 5: Revisions to population data, in our Population estimates for England and Wales: mid-2024 bulletin.

The SPD V5.1 is produced using the Demographic Index (DI). This is a database that matches people across multiple different administrative data sources. The DI is not adjusted to reflect the usual resident population. All datasets that contain information about people are referenced to the DI and given an ONS identifier to facilitate linkage between sources. Information about the further research planned to explore the quality of the DI is available in the Methodological evaluation and quality assessment of the Reference Data Management Framework (RDMF) report (PDF, 295 KB) from the UK Statistics Authority (UKSA).

To produce SPD V5.1, we apply a set of rules to the DI that approximate the resident population. This means that we can link both the Census 2021 veterans records and the SLD veterans records to SPD V5.1. Census is linked directly using ONS IDs. The SLD is linked to the DI using NHS numbers to assign ONS ID to those records. This enables linkage to SPD V5.1.

Figure 1: Linkage methodology for producing UK armed forces veteran statistics for England and Wales

Source: Office for National Statistics

Download this image Figure 1: Linkage methodology for producing UK armed forces veteran statistics for England and Wales

.png (36.7 kB)

Undercoverage and overcoverage

There is no adjustment made for the undercoverage and overcoverage of veterans in the dataset. Undercoverage is when our data sources are not picking up people that should be included our target population. This can occur in two scenarios.

Firstly, some individuals are not captured at all in the SPD V5.1 for 2023 (for example, because they did not have enough interactions with administrative datasets) or it is not possible to link their administrative data records correctly. These groups could include:

• people who were recorded as veterans in Census 2021 

• people who were veterans on Census Day in 2021 but who did not respond to Census 2021 stating that they were a veteran (either because they did not answer the veterans question or did not respond to the Census at all) 

• people that have become veterans since Census 2021 

Secondly, there were some individuals who were included in SPD V5.1 for 2023 but met one or more conditions. These were that:

• they did not respond to Census 2021 stating that they were a veteran, or did not respond to the census at all, so we do not know whether they were a veteran based on census data

• their census record did not link to the SPD, or there were linkage issues around their census record 

• they left service at a time when the SLD did not exist, so would not have linked to the SPD (this will affect those aged 64 years and over in 2023, who did not appear as a veteran on Census 2021) 

• they left service when the SLD existed, but we have not been able to link them to the SPD (either because of missing or poor-quality NHS number data, or because of another linkage issue); this is more likely to affect some age groups than others, but there are still some very recent records with linkage issues

Finally, overcoverage may also have occurred where an individual was incorrectly identified as a veteran according to the harmonised standard. When the ONS receive the SLD, currently serving personnel have been removed from the dataset by the MoD using payroll data. This means that if someone confirmed they were a veteran in the census but has since returned to active service, this method would still flag them as a veteran. This is unlikely to affect those aged 64 years and over, as a record over this age that only appears on Census 2021 is likely to have left service before the SLD's earliest coverage year (1975).

Linkage Analysis

There were 1,534,685 veterans in the 2023 SLD, of which 429,140 (28%) had valid NHS numbers, meaning they were eligible for linkage to the SPD V5.1. Of these records, 424,125 (99%) linked to the DI and 335,345 (78%) linked to the SPD V5.1. There were 1,761,840 veterans in Census 2021, of which 1,483,275 (84%) linked to the SPD V5.1.

There are several factors that explain the low overall linkage rate for the SLD to the SPD V5.1, using the DI. These factors are that:

• the DI only consists of administrative data from 2016 onwards; anyone who left service from 1975 and died before 2016 will not be in the usual resident population and therefore will not link

• the SLD includes veterans who have died, these records would not link to the SPD V5.1, which is designed to reflect the usual resident population in mid-2023

• the NHS number quality is poorer on the SLD for records with earlier service exit dates

The strength of Census 2021 data is that it links veterans from 2021 or earlier, while the SLD's strength is supplementing veterans post-2021, although all years of SLD data are used in the analysis. When isolating the SLD to records with a service exit date from mid-2021 onwards, we saw much improved linkage rates. There were 37,710 SLD records with a service exit date from mid-2021 onwards and of this sub-population there were 33,850 (90%) SLD veterans with valid NHS numbers, meaning they were eligible for linkage to the SPD V5.1. Of these records, 29,510 (87.2%) linked successfully to the SPD V5.1.

Veterans identified for each source

When investigating which sources veterans were identified from, we saw that the majority were found in Census 2021. There was an overall increase in veterans identified by the Service Leavers Database (SLD) of 11,100, compared with the 2021 feasibility research data published in 2023. Overall, there was a decrease of 148,005 in the total number of veterans identified compared with the 2021 feasibility research data.

This decrease is likely because of deaths in the older veteran population, specifically veterans who only responded to the census and who were not on the SLD, which would explain the large decrease in "census-only" veterans. This conclusion is supported by comparing the number of census-only veterans aged 75 years and over between the two studies, as there was a decrease of 79,670 (the largest of all age groups).

The Statistical Population Dataset version 5.1 (SPD V5.1) also contains different income and benefits data sources compared with SPD V4.2, which was used in the 2021 feasibility research. This is particularly true of state pension data, meaning there could be older veterans who are not included because of these coverage differences. Figure 4 shows the overall age distribution of UK armed forces veterans in the SPD.

Veterans identified for each source by age

Census 2021 identified a much larger proportion of veterans aged 65 years and over compared with the SLD, with the SLD providing more effective coverage of the younger age groups. This census bias is because the SLD only covers veterans from 1975 onwards, coupled with poor NHS number quality for more historic SLD records.

To compare age distributions more effectively between sources, we investigated records for veterans aged between 16 and 60 years. This was so we could look at ages more likely to include both census and SLD veterans, because the SLD coverage does not begin until 1975. SLD veterans were more likely to be aged between 32 and 43 years, with a peak at 37 and 38 years (both 4.1%). Census veterans were more likely to be aged 51 years and older with a peak at 60 years (5.7%).

This shows us that the SLD and Census 2021 were providing coverage of different types of veteran populations. The SLD was more likely to capture younger veterans leaving post-census, and the census was more likely to capture older veterans who were not in the SLD because of historic quality issues. This shows the importance of the combined census and administrative data method, using both sources to build a fuller picture of the veteran population.

Veterans identified for each source by age and sex

Most veterans identified in each source were male which aligns with 2021 figures from Census 2021 (which records 86% of veterans as male). A very small proportion of female veterans was identified through the SLD, this proportion decreased even further for veterans aged 45 years and over. The SLD captured male veterans aged up to 76 years, with the majority identified at younger ages. Both male and female veterans were exclusively identified using census data from the age of 77 years. For Census 2021 only, those aged 16 years and over on Census Day were asked the veteran question. This means that records were only identified as a veteran in Census 2021 from the age of 18, as we aged the population on from 2021 to produce 2023 statistics. We rely on the SLD to fill this gap.

Figure 4 also shows that 16% of 18-year-old and 25% of 19-year-old veterans were female veterans who were identified on the census but not on the SLD. These proportions of young female veterans compared with young male veterans are higher than might be expected when comparing with trends from the UK armed forces biannual diversity statistics from the Ministry of Defence (MoD) for outflow from UK regulars. While this age group is a small proportion of the overall veteran population, and there are differences in data collection and definitions between these sources, this is a trend we will investigate further in future research.

Our method identified that 3.3% of the Statistical Population Dataset version 5.1 (SPD V5.1) in England and 4.1% in Wales had previously served in the UK armed forces, as of 30 June 2023.

In the SPD V5.1, 80.3% of individuals in England were identified as not having previously served in the UK armed forces, according to Census 2021, and did not link to a Service Leavers Database (SLD) record. For Wales, this figure was 80.7%.

We were not able to establish whether an individual was a veteran or not for 16.4% of individuals resident in England and 15.2% of individuals resident in Wales in SPD V5.1. Individuals could be in this group for one of several reasons, including:

• they had an NHS number in the SPD V5.1, but did not link to Census 2021 or the SLD; this comprised 11.3% of the SPD V5.1 for England, and 9.2% for Wales 
• the veteran information was missing on Census 2021, and these individuals also did not link to the SLD (4.9% of the SPD V5.1 for England, and 5.8% for Wales); census records that were either entirely imputed or only had item imputation applied to the veteran question were included in the population with missing Census information
• the individual did not have an NHS number on the SPD V5.1; this comprised 0.2% of the SPD V5.1 for England, and 0.2% for Wales

Coverage statistics of veteran status

Coverage statistics of veteran status by age

The proportion of veteran records in the SPD V5.1 was largest for those aged 84 years and over. This is likely because of National Service, which operated between 1939 and 1960. The proportion of SPD V5.1 records that did not link to Census 2021 or the SLD varied by age, with the largest proportion being between 21 and 25 years.

The proportion of records that were unknown because of missing Census 2021 or SLD data was skewed considerably at 16 and 17 years old. This is because those who were 16 in 2023 would have been 14 in 2021 and so would not have answered the Census veterans question, which was only asked of those aged 16 and over as of Census Day. In future versions of the methodology, we could assume that all those born after 21 March 2005 (who would not have been 16 years old on Census Day) are not veterans, unless they have linked to an SLD record. Further information on this can be found in Section 7: Data Sources and Quality.

Coverage statistics of veteran status by age and sex

Most veterans in the SPD V5.1 population are male (87%), which is in line with Census 2021 figures (which state that 86% of veterans were male). We can also see that the proportion of males is higher for the unknown category of the SPD V5.1, which includes those with missing NHS numbers. This is potentially a result of young males being less likely to update their health information.

When looking at age and sex together, we see that the highest proportion of female veterans are found in those aged 90 years and over (5%). This is likely because of the 1941 National Service Act which extended conscription to include females. After 1945, National Service did not extend to females, however, it continued for males until 1960.

Coverage statistics of veteran status by local authority

The local authorities with the highest proportions of veterans in the population are:

• Gosport (12.1%)

• North Kesteven (10.3%)

• Fareham (8.9%)

• Plymouth (7.9%)

• Wiltshire (7.3%)

This broadly aligns with findings from Census 2021; however, this is to be expected given the large proportion of veterans identified through the census compared with the SLD. Though the similarity to the census provides assurance of the method, regardless of source breakdown.

When looking at local authorities by source of identification, those with the highest proportions of SLD veterans are:

• North Kesteven (37.4%)

• Wiltshire (33.6%)

• Plymouth (32.2%)

• Darlington (32%)

• Gosport (31.8%)

This shows that the SLD is following similar trends to the census regarding the local authorities where veterans are living.

Figures for all local authorities can be found in the accompanying dataset.

Population Base

The 2023 Statistical Population Dataset version 5.1 (SPD V5.1) for England and Wales, filtered to those aged 16 years and over.

UK armed forces veteran

People who have previously served in the UK armed forces. This includes those who have served for at least one day in the UK armed forces, either as regular, reserves, or merchant mariners who have seen duty on legally defined military operations. Further information is provided in the Harmonised standard for previous UK armed forces service (under development).

Usually resident population

We are currently adopting the UN definition of "usually resident", which refers to the place at which a person has lived continuously for at least 12 months, not including temporary absences for holidays or work assignments, or intends to live for at least 12 months (United Nations, 2008).

Data sources

We used the 2023 Statistical Population Dataset version 5.1 (SPD V5.1) for England and Wales, filtered to those aged 16 years and over, as the population base for our feasibility research on producing veteran statistics. The quality of the population base will impact the quality of our feasibility research on producing veteran statistics. 

Veterans in the feasibility research have been identified using two data sources.

Service Leavers Database, 1 January 1975 to 30 June 2023

This database provides information for service personnel who have left the UK armed forces, sourced from legacy personnel systems and the current system, Joint Personnel Administration (JPA). We receive a subset of variables from the Service Leavers Database (SLD) for data back to 1975 from the Ministry of Defence (MoD).

Census 2021

In 2021, a new question was added to the census that asked those aged 16 and over as of Census Day (21 March 2021): "Have you previously served in the UK Armed Forces?" with response options of "Yes, previously served in Regular Armed Forces", "Yes, previously served in Reserve Armed Forces" and "no".

Unknown veteran status at 16 and 17 Years

The current methodology will always show large proportions of 16- and 17-year-olds with "unknown" veteran status. We could assume those born after 21 March 2005 (and who were therefore not 16 on Census Day) were not veterans, unless there is an SLD record for them. Some could be missing because of linkage issues, but there is no way of distinguishing at this stage, given there is no census record to confirm they were not veterans.

We will build on our findings to aid our understanding of the quality considerations we have identified, particularly relating to accuracy and coherence. Potential research could include:

• approximating the maximum number of possible missed veterans on the Statistical Population Dataset version 5.1 (SPD V5.1) by applying mortality rates to the veterans that do not link between the DI and SPD V5.1

• comparing veteran aggregates from SPD V5.1 for 2023 with rolled-forward Census 2021 estimates, adjusted for mortality, emigration and immigration, including movement into England and Wales from other UK countries and abroad

• exploring methods to better understand the characteristics of the SPD V5.1 population that do not link to Census or SLD records, given that different groups within the population will have different likelihoods of being a veteran; as we move away from Census 2021, more records may not link to either source, because they are not a veteran and did not respond to Census 2021

We welcome feedback on the suitability of the data, research and planned future developments for this method. This will allow the Office for National Statistics (ONS) to meet growing user need for official veteran statistics and to fulfil its responsibility towards the Armed Forces Covenant. Please email your feedback to pop.info@ons.gov.uk.