Official Statistics

Background quality report: MOD supported employment estimates 2022/23

Published 19 September 2024

1. Contact Details

The Analysis Directorate welcomes feedback on our statistical products. If you have any comments or questions about this publication, or about our statistics in general, you can contact us as follows:

Analysis-Expenditure Head of Branch

Telephone: 030 015 86554

Email: Analysis-Expenditure-PQ-FOI@mod.gov.uk

If you require information which is not available within this or other available publications, you may wish to submit a Request for Information to the Ministry of Defence under the Freedom of Information Act 2000.

Analysis Directorate (Analysis-Expenditure)
Ministry of Defence
Oak 0 West, #6028
MOD Abbey Wood North
Bristol
BS34 8QW

For general MOD enquiries, please call: 020 7218 9000

Find out about call charges

2. Introduction

2.1 Overview

These statistics present information on employment supported by MOD regional expenditure with UK industry in 2022/23, with some comparisons made to previous financial years. Using expenditure data in conjunction with employment and turnover data from the Business Register and Employment Survey (BRES), and Annual Business Survey (ABS) respectively, we have calculated an estimate of the number of direct jobs supported by MOD expenditure with UK industry. MOD expenditure data can also be used as an input to indicate MOD final demand for goods and services. Using this with Office for National Statistics (ONS) Input-Output Analytical Tables (IOATs) and Supply-Use Tables (SUTs), we can estimate the number of indirect jobs supported.

For both employment types, the estimated number of jobs are shown broken down by industry group. Additionally, for direct jobs this is also presented by ITL Level 1 regions, referred to as regions in text, and as the number of direct jobs supported by MOD expenditure per 100,000 full-time equivalent (FTE) people employed in the region. This makes the figures between regions more directly comparable to one another by adjusting for varying regional populations and different employment levels.

2.2 New Format

Since 2013/14, job statistics were published in MOD Regional Expenditure with UK Industry and Supported Employment statistics. In 2023, the decision was made to split the expenditure and jobs estimates into separate statistics to improve on timeliness and quality of the expenditure statistics.The expenditure estimates for 2022/23 were published in February 2024 and can be found in the MOD Regional Expenditure with UK industry bulletin.

The methodology used to produce the regional expenditure estimates has been reviewed for the 2022/23 publication. As a result, revisions have been made to the estimates of regional expenditure and jobs supported by this expenditure for 2020/21 and 2021/22 to allow for consistent comparisons across years.

Further revisions have been made to the jobs estimates by transitioning to more up to date versions of the ONS Input-Output Analytical Tables (IOATs) and Supply-Use Tables (SUTs), which are used in the estimation process:

• 2018/19 jobs were estimated using 2018 IOATs and SUTs
• 2019/20 through to 2022/23 jobs were estimated using the 2019 IOAT and SUTs.

The ONS revised their methods for producing the IOATs from 2016 onwards, which has had an impact on our estimates of indirect jobs. This has led to some large changes in our estimates in the revised years in certain sectors, including the ‘Manufacture of Weapons and Ammunition’ sector, compared to our estimates using the 2015 IOATs.

Whilst the 2020 version of the IOATs and SUTs have been published, the decision was made not to use them in our calculations as the ONS advised that the tables would be impacted by the COVID-19 pandemic. It was deemed it would not be appropriate to estimate the MOD supported jobs in non-COVID years with those tables, so the 2019 IOAT and SUTs were used for the years 2019/20 through to 2022/23 for consistency. Once future years of the IOAT and SUTs are available, we will be able to apply the relevant year to each year of our job estimates.

Overall, the revisions have resulted in the following percentage increases in the estimates of the total jobs supported by MOD expenditure with UK industry: 5% in 2018/19, 2% in 2019/20, 4% in 2020/21 and 8% in 2021/22.

3. Statistical Processing

This section sets out the data requirements and processes used to create the tables and charts in the core bulletin and supporting data tables. It also discusses the assumptions made as well as a number of limitations.

3.1 Source Data

As well as using MOD expenditure data identified by MOD’s Contracting, Purchasing and Finance (CP&F) system and information from individual MOD project teams, considerable data input for the estimation of jobs relies upon releases from the ONS.

• Supply-Use Tables (SUTs) are produced annually to show estimates of industry inputs and outputs, product supply and demand, and gross value added for the UK. SUTs for 1997 to 2021 are consistent with the UK National Accounts 2022 Blue Book.
• Input-Output Analytical Tables (IOATs) are produced semi-regularly and are derived from the SUTs. They highlight how products are used to produce further products and satisfy final demand across ONS industry codes.
• NOMIS annual data on regional employee and employment counts.
• Data on UK employment and turnover by MOD Standard Industrial Classification (SIC) group is produced by the ONS as a subset of the Business Register and Employment Survey (BRES) and Annual Business Survey (ABS).

3.2 Data Compilation: Direct Jobs

Using employment data from the BRES and turnover data from the ABS we can calculate turnover per FTE employment for each financial year. By dividing our figures for MOD expenditure with UK industry by these figures we can estimate how many direct jobs this expenditure supports in the UK. Again, these figures are presented broken down by both region and industry group.

We use further employment data from BRES and NOMIS to calculate how many jobs MOD expenditure supports for every 100,000 people in FTE employment in the region. This takes into account both the population of the area and the number of people in employment and therefore, like the per person expenditure figures, makes it easier to make comparisons across the different regions.

3.3 Data Compilation: Indirect Jobs

To estimate the number of indirect jobs supported in the UK by MOD expenditure, we use direct MOD expenditure with UK industry as a measure of MOD’s demand for products and services across the UK economy. By aggregating the ONS Input-Output Table from the IOATs to match the SIC groups used by MOD, we can use this table to show total UK wide demand and output (including intermediate products) arising from this initial MOD demand. Subtracting MOD final demand from this leaves just the intermediate demand (i.e. that which occurs throughout the supply chain). An estimate of UK output per FTE employment is calculated using the ONS Supply of Products Table in the SUTs and BRES employment data. The amount of UK output generated from MOD intermediate demand is then divided by output per FTE employment to determine the number of indirect jobs supported by MOD expenditure. Due to the method employed in their derivation, as well as showing the overall total, these estimates can be presented by industry group.

Figure 1: Overview of Methodological Process

Description of Figure 1: Flowchart providing an overview of the indirect jobs’ methodology. To output indirect jobs, data inputs are MOD expenditure with each industry, ONS Domestic Use Table, ONS Supply of Products Table and Business Register and Employment Survey data.

This approach has been applied over that of employment multipliers since it is consistent with the method of estimating direct jobs whereby jobs are estimated as a ratio of industry output to employment. As such, the data required for MOD expenditure and UK employment is readily available to the Analysis Directorate. The applied method also allows for the exclusion of employment within the public administration and defence sector. The approach is broadly consistent with the way estimates were previously produced by UKDS.

The output from calculating MOD spending using the aforementioned methodology is a matrix of expenditure split across SIC groups and UK regions. Since there are no regional SUT tables, the regional split is ignored for the estimation of indirect jobs and so the output expenditure is consolidated to become a single vector of MOD final demand split by MOD SIC groups 1 to 52.

The ONS Domestic Use Table frequently appears in later releases of the UK Input-Output Analytical Tables as the IOT table, it being the main Input-Output Table. It shows which products (rows) go to produce other products (columns) as intermediate production.

Table 1: Format of IOT Table (Domestic Use, Basic Prices, Product by Product)

Description of Table 1: Table illustrating the generalised layout of the Office for National Statistics’ Input-Output Table, also known as the Domestic Use Table, in basic prices and product by product.

At the base of the table can be found the total output for each product after allowing for taxes, subsidies, etc. Owing to the nature of the tables and their initial construction, they are balanced, and the total output equals total demand.

Consider a reduced example from the 2019 Input-Output Analytical Tables. Table 2 shows that the product of ‘Air and spacecraft, and related machinery’ that goes into the product of ‘Weapons and ammunition’ is £4.9 million. The final product will then have some element of value added through taxes, compensation of employees, etc. as shown at the bottom of Table 1 before its total output is found.

Table 2: Reduced Section of 2019 Input-Output Domestic Use Table at Basic Prices, Product by Product

		ONS Product Group
ONS Product SIC Code	ONS Product Description	25.4: Weapons and ammunition (£ millions)
27	Electrical equipment	70.7
28	Machinery and equipment n.e.c.	129.9
29	Motor vehicles, trailers and semi-trailers	0.1
30.1	Ships and boats	0.0
30.3	Air and spacecraft, and related machinery	4.9

Currently there is a mismatch between the ONS SIC codes and MOD SIC groups. Starting with the relevant IOT showing domestic use in basic prices, product by product, we can aggregate the groupings by simply summing rows to go from the 100+ ONS SIC codes to MOD’s 52 SIC groups. Where SIC codes fall into multiple SIC groups a decision rule is applied to accommodate the overlaps whereby amounts are apportioned equally. This aggregation keeps all row and column totals intact but allows us to view demand according to MOD’s own SIC groupings.

As it stands, Government expenditure is included within the intermediate demand section of the Domestic Use Table (Table 1). Since we are not interested in subsequent jobs in the public administration and defence sector, any intermediate demand against SIC group 48 is nilled out in both row and column. This is considered consistent with the estimation of direct jobs where expenditure identified as being with other government departments is removed.

Dividing each cell of the aggregated use table by the column total produces a matrix of coefficients (also known as the A matrix). This new matrix shows how much of each product is required across all SIC groups when producing one final unit of a certain product. For example, in producing weapons and ammunition you are going to need elements of metal work and electrical components among others, as well as further work in the weapons and ammunition sector itself.

Table 3: Reduced Example of the A Matrix (Matrix of Coefficients)

	SIC Group 1	SIC Group 2	SIC Group 3
SIC Group 1	0.14	0.00	0.00
SIC Group 2	0.00	0.00	0.00
SIC Group 3	0.00	0.00	0.05

Table 3 shows a reduced example of a constructed A matrix and the interactions between three products. It shows that to produce one unit of product from SIC group 1 it would require 0.14 units of SIC group 1, 0.00 of SIC group 2 and 0.00 of SIC group 3. However, since SIC group 2 also has interactions with SIC groups 1, 2 and 3, these requirements would also need to be calculated and factored in. These interactions would go on infinitely.

Instead of doing this manually for each individual interaction we turn to matrix calculations to multiply the A matrix by our demand for each step.

Let f be the initial vector of final demand, where for example,

Description of mathematical notation: denotes the initial vector of final demand where f equals a three by one matrix with values one, zero, zero.

Then A x f gives the secondary product requirements, A x A x f gives the tertiary product requirements, and so on. These interactions are repeated infinitely which can be written as,

f + Af + A2f + A3f + … . (†)

Let I be the identity matrix. The above is then equivalent to

(I + A + A2 + A3 + …)f

We want to reduce this infinite series into a finite form which we can calculate explicitly. For brevity, let R = I + A + A2 + A3 + … . Then, since

I + A + A2 + A3 + … = I + A(I + A + A2 + … ),

we have R = I + AR.

We can now rearrange this as follows:

Description of mathematical notation: shows how the equation, R equals I plus A times R, is equivalent to, R equals the inverse of I minus A, i.e. the Leontief Inverse matrix.

Therefore, we may reduce the infinite series I + A + A2 + … to the Leontief Inverse Matrix, (I - A)-1, which shows how much of each industry’s output is needed to produce one unit of a given industry’s output. For any vector of final demand, total demand for the product is found by multiplying it by the Leontief Inverse.

We can therefore write (†) as (I - A)-1f and calculate the total demand for the above example:

Description of mathematical notation: calculates total demand. The A matrix is subtracted from the identity matrix, and the inverse taken. This is then multiplied by the initial vector of final demand to produce a three by one matrix for total demand.

So, to produce one full unit of product from SIC group 1 we would require 1.18 units of total product of SIC group 1, 0 of SIC group 2 and 0 of SIC group 3.

To create a MOD version of the Leontief matrix we take the aggregated domestic use table and then perform these matrix transformations. This therefore produces a Leontief matrix of 52 by 52 SIC groups which then directly matches up to MOD’s breakdown of SIC groups used to report on estimated job numbers.

MOD total demand across SIC groups is then calculated by multiplying the Leontief matrix by the vector of final demand, as per the example above. Importantly, to account for the jobs supported by the direct expenditure across SIC groups, we only calculate indirect jobs on the difference between MOD total demand and the final demand vector. This is known as the intermediate demand, that which occurs throughout the supply chain.

Total Demand – Final Demand Vector = Intermediate Demand

Description of mathematical notation: calculates intermediate demand. Subtracting final demand from total demand yields intermediate demand. In our example, this is a three by one matrix with values 0.18, 0 and 0.

In order to determine employment numbers, we start by calculating output at basic prices per FTE employment. The SUTs include a Supply of Products table which features data on total domestic output of products at basic prices. Being in basic prices ensures it is consistent with that of the IOT table.

A MOD version of the SUT’s Supply of Products table based on MOD’s 52 SIC groups can be created by aggregating SIC groupings by the same method applied in compiling the MOD’s IOT. This returns total domestic output of products at basic prices broken down according to MOD SIC groups. Using the employment data gathered in the BRES, dividing output by employment here provides the output at basic prices per FTE employment. The number of indirect jobs can then be estimated by dividing the intermediate demand (MOD total demand minus the MOD vector of final demand) by these output at basic prices per FTE employment figures.

Table 4: Reduced Example of Indirect Jobs Estimation

	Total Domestic Output of Products at Basic Prices (£)	Total FTE Employment	Output at Basic Prices per FTE Employment (£ per FTE)	Intermediate Demand	Indirect Jobs Estimate
SIC Group 1	32,000	50	600	0.10	0.00016
SIC Group 2	200	1	300	0.00	0.00000
SIC Group 3	16,000	30	500	0.00	0.00000

So, with our given final demand vector plus the example output and employment data in Table 4, for each £1 increase in final demand of SIC group 1, intermediate demand would support an estimated 0.00016 indirect FTE jobs in SIC group 1, 0.00000 in SIC group 2 and 0.00000 in SIC group 3.

For further detail on this methodology for indirect jobs, a full worked example can be found in Annex A.

3.4 Assumptions and Limitations: Direct Jobs

• MOD’s financial year data can be matched to ONS calendar year data.
• The number of direct jobs supported is only an estimated figure. There is no concrete way of counting the number of jobs supported by MOD expenditure with UK industry. Instead we have taken the turnover per FTE employment in each industry group and divided our regional expenditure figures by this. This gives an estimate of how many jobs MOD expenditure supports in the UK.
• For some SIC groups either turnover data or employment data was not available due to suppression. In these cases, data was estimated based on other available financial years (if similar SIC groups had consistent data across financial years). If a lot of variation was seen across financial years in a similar SIC group then the average turnover per FTE employment was applied here instead. Please contact us if you need information on which SIC groups were affected by this.
• BRES data was not available at the level we required it for Northern Ireland and therefore we had to use published figures from the UK Business Register and Employment Survey directly from the ONS website.

3.5 Assumptions and Limitations: Indirect Jobs

• Detailed IOATs containing the necessary product breakdown used in the estimation process for indirect jobs were only produced for 2010 and 2015 and then from 2018 onwards. The IOATs used for each financial year are reported in the supporting tables of these statistics and are deemed most appropriate to use but will not align directly with the reporting years. The 2020 version of the IOATs is currently available but was however not used as those estimates would have been impacted by the COVID-19 pandemic.
• SUTs were used alongside the IOATs of the same year. It must be assumed that total domestic output of products does not significantly change year-on-year.
• Defence output from a particular SIC product group will have similar characteristics as total output for the sector as a whole.
• The MOD vector of final demand is in current prices whereas the ONS analytical tables are in basic prices so it must be assessed whether they are compatible. Basic prices are the amounts received by the goods or services produced minus any taxes and subsidies on products. The expenditure in the MOD vector of final demand does exclude VAT but in order to match to basic prices it is assumed that other taxes paid by MOD are relatively small.
• Not every ONS SIC group falls into a single MOD SIC group, some overlap across multiple groups. For example, ONS SIC 26 for ‘Computer, electronic and optical products’ lies in MOD SIC groups 22 for ‘Office machinery and computers’, 24 for ‘Electronics’ and 25 for ‘Instrument engineering’. Decision rules must be applied to determine how these overlaps should be grouped. Currently these splits are apportioned equally.
• IOTs are prepared as product by product while SIC groups used are, by definition, industry based. There is therefore a mismatch between the data on a product basis then being used by industry to calculate employment.
• No reliable regional split can currently be given for estimates on indirect jobs as there are no regional IOATs. It takes considerable time and effort for the ONS to produce UK tables so producing regional ones as well is not feasible. So, as it stands, indirect jobs are reported for the whole of the UK rather than by region as this would require the large assumption that each region has the exact same make up for product supply, use and demand.

4. Quality Management

4.1 Quality Assurance

The general production process is enhanced by using an open-source programming language to automate large parts of the data processing. Numerous quality assurance measures are programmed into the data pipeline, reducing the scope for human error that was more likely to occur when doing a large volume of manual data. In addition, production time is reduced allowing more time to improve on quality and accuracy of the publication and can be easily reproduced.

4.2 Quality Assessment

The MOD’s quality management process for Official Statistics consists of three elements:

1. Regularly monitoring and assessing quality risk via an annual assessment.

2. Providing a mechanism for reporting and reviewing revisions/corrections to Official Statistics.

3. Ensuring BQRs are publishing alongside reports and are updated regularly.

5. Relevance

Statistics on the number of jobs that supported by MOD expenditure with UK industry were published annually up until 2009 in the UKDS publication. This publication was re-introduced in response to a request from the Secretary of State for Defence to begin producing these statistics again, which was supported by the Permanent Under Secretary in March 2016.

Due to increased demand to once again produce figures on indirect jobs, an Experimental Statistic was released by the Analysis Directorate containing these estimates in the Indirect Jobs bulletin. These figures were subsequently subsumed into the Regional Expenditure bulletin until 2021/22 and are now published as part of the current Supported Employment by MOD Expenditure with UK Industry bulletin.

The information contained in this bulletin has potential for a wide range of users including the media, politicians, policy professionals and the general public. In particular, these statistics aid the measurement and monitoring of various regional expenditure targets throughout the MOD and wider government. Regional expenditure statistics are often used in parliamentary debate, in answering parliamentary question and requests from industry on MOD spending.

The MOD invites users to provide further feedback to the statistical output teams on any of their publications or reports and continues to assess the scope of publications to meet user needs.

6. Accuracy

Accuracy details of some of the key outputs of the regional employment statistics are detailed below.

Industry groups - MOD industry grouping are based on SIC codes assigned to contracts, for most contracts a single SIC group code is assigned to each contract, however, each contract will likely span multiple different industries groups. The vast number of contracts let each year means it is not possible to assess each contract individually therefore, where possible we contact the project teams who own the larger contracts to obtain more detailed SIC data. From 2019/20, an upgrade to the CP&F system has provided Commercial Officers the option to add in multiple SIC details which allows for an improved allocation of expenditure to work type.

7. Timeliness and punctuality

7.1 Timeliness

Publication of these statistics is reliant on supporting data from the ONS that enables us to produce our estimates. Data on UK turnover by industry groups is typically not available until May of the following year which is then further impacted by the additional analysis and quality assurance processes required before publication. As a result, there is significant delay between the reporting year and publication.

7.2 Punctuality

The release date for this publication was pre-announced on the MOD’s Calendar of Upcoming Releases section of GOV.UK.

8. Coherence and comparability

8.1 Data Comparability

By breaking supported employment down into ITL Level 1 Regions, comparisons can be made about different areas of the UK. The inclusion of jobs supported per 100,000 people in FTE employment also makes the statistics more comparable across different regions. Without these measures, differences in employment levels of the regions could make it difficult to compare jobs supported.

9. Accessibility and clarity

The statistical bulletin can be accessed on the GOV.UK website where it is available to download in HTML format. Its release is noted in the Finance and Economics section of MOD’s list of national and official statistics by topic and can also be found by using an internet search engine.

Visualisations have been chosen to best display patterns and trends within the data. Terms used in the commentary are defined within the glossary of the HTML bulletin. All tables and data behind any graph or chart in the report are available as accessible Open Data Source (ODS) tables.

Figures within the bulletin and ODS tables are often rounded to aid with clarity. In these instances, we have followed the Ministry of Defence Rounding Policy.

Should you have any feedback on the accessibility of any part of the bulletin or accompanying data tables then the Analysis Directorate encourages you to get in touch via any of the means noted in the Contact Details section.

10. Cost and Respondent Burden

In producing these statistics, our main data sources are administrative data which are used for many purposes. On occasion the data has required manual cleansing due to missing or incorrect information following the introduction of CP&F. However, this cleansing cannot be avoided as it is the only way to improve data quality to the standard required for this publication. Going forward we intend to use wider education methods in consultation with Commercial Managers to improve data quality.

ABS and BRES data according to MOD SIC groupings used in the turnover per FTE employment calculations are extracted on our behalf by the ONS. This saves time as they are experts on this data and can produce the figures much faster than we can.

11. Confidentiality and Security

11.1 Confidentiality - Policy

In producing these statistics, we adhere to the MOD Analysis Directorate Confidentiality Policy. A disclosure policy for commercial data has been agreed and a process now exists for deciding on the release of data that is consistent with the Transparency Agenda and the existing rules relating to the answering of Freedom of Information requests.

We adhere to the principles and protocols laid out in the Code of Practice for Official Statistics and comply with pre-release access arrangements. The MOD Statistics Pre-Release Access Lists are available on the GOV.UK website.

11.2 Confidentiality - Data Treatment

The Analysis Directorate maintains good links with policy colleagues to ensure that these statistics are understood and to prevent misuse. We regularly review our commentary and visualisations to ensure the data is presented in the best way possible.

11.3 Security

The team operate a secure environment for the storage of sensitive commercial data and other linked data. All data used in this bulletin is stored and managed securely on an internal SQL server.