Incorporating the Annual Survey of Goods and Services into the UK National Accounts, Blue Book: 2026

Every year, the Office for National Statistics (ONS) updates the sources and methods for the UK National Accounts and publishes the latest estimates (including revisions to past periods) in our annual UK National Accounts, the Blue Book compendium. 

Our recent Proposed changes to be implemented in Blue Book and Pink Book: 2026 article described proposed changes to the UK National Accounts, and balance of payments, that will be made in our Blue Book 2026 and Pink Book 2026 releases.

One of the main methodological improvements we will implement is integration, in the production measure of gross domestic product (GDP) of data from the Annual Survey of Goods and Services (ASGS).

The UK economy has a large proportion of activity in the service sector. Historically, readily available data to quantify products that businesses in the service sector produce has been limited. For the manufacturing sector, we use our UK Manufacturers' Sales by Product Survey (PRODCOM). To improve the understanding of the output of the service sector at a disaggregated product level, we developed a comprehensive survey of services industries, the ASGS.

As with PRODCOM data in the manufacturing sector, the ASGS data will now be used to allocate industry level output derived from the Annual Business Survey (ABS) to individual products. This represents an important quality improvement, fully replacing product patterns derived from our previous Services Turnover Survey.

These patterns were partially aligned to early iterations of the ASGS during the implementation of double deflation. For the 2025 data release they were further refined in targeted cases, for example the research and development (R&D) industry. The 2026 data release will include the full range of the ASGS data, following a substantial and detailed quality assurance process.

The survey data are available from 2016 to 2024 and cover non-financial businesses. However, 2016 data are not used in the national accounts because of quality issues with the granular data required for the national accounts. The ASGS data are back-cast from 2016 to 2007 to avoid a discontinuity in the GDP data.

The ASGS does not change our starting estimate of the total output of industries in the non-financial business economy; this comes from the ABS. However, the improved understanding of the range of service products led us to re-evaluate how we reconciled data in our supply and use framework. Therefore, the ASGS will have some effect on current price (nominal) headline GDP. Some additional impact on real GDP (chain volume measures) is expected, as annual double deflation methods are based on products rather than industries.

This article explores some of the methodological considerations behind the integration of the ASGS data in the UK National Accounts. It does not consider detailed data impacts. This will form part of later articles, published in August 2026, which will cover all the implemented changes in the 2026 Blue Book.

An overview of the GDP annual method

Gross domestic product (GDP) is estimated using three approaches:

• the production approach

• the income approach

• the expenditure approach

In theory, the three different approaches should produce the same result. However, as they are based on different surveys and administrative data sources, they all produce results that are subject to different elements of uncertainty.

Each of the three approaches are separately compiled and confronted in the annual supply and use framework to derive a single estimate of the current price level of GDP. Following initial balancing at current price, volume GDP is estimated using the double deflation approach.

The UK supply and use tables show the:

• composition and value of goods and services entering final demand

• outputs and incomes generated in the production process

• intermediate transactions that form inputs into these processes

The tables are constructed to show a balanced and complete picture of the flows of products in the economy; they illustrate the relationships between producers and consumers of goods and services. They achieve consistency in the economic accounts' aggregates by linking the components of value added, inputs, outputs, and final demand.

The balancing process takes in to account the relative strengths and weaknesses of each component in a granular matrix of industries and products. Many of the components are further subdivided into sectors, for example:

• the financial corporations' sector

• the government sector

• the household sector

These sectors are groupings of units (the individual households or legal entities, such as companies), that participate in the economy.

The integration of the Annual Survey of Goods and Services (ASGS) data in GDP directly affects one specific element of the supply and use balancing framework. That is, the product breakdown of market output by the private non-financial corporations' sector within the production approach. 

Calculating annual GDP by the production approach

The production approach to GDP is based on the sum of the value added created through the production of goods and services in the economy (our production or output as an economy). This approach can be used to show how much different industries (for example, manufacture of textiles or air transport) contribute to the economy.

Gross value added (GVA) for each industry is calculated as output (P1), less intermediate consumption (P2), and is transformed to GDP by adding taxes on products and removing subsidies on products.

Output comprises market output (P11), output for own final use (P12), and non-market output (P13). Within the industry framework, GVA is generated by each of these sectors:

• private non-financial corporations (PNFCs)

• public corporations (PCs)

• financial corporations

• government

• households

• non-profit institutions serving households (NPISH)

These sectors rely on different sources to calculate industry level output and intermediate consumption, and to imply product breakdowns.

Gross value added by the private non-financial corporations sector

For the private non-financial corporations (PNFCs) sector, market output for each industry is largely determined by data returned by businesses selected for the Annual Business Survey (ABS). For some industries alternative sources are used, for example, the agriculture industries use Department for the Environment, Food, and Rural Affairs (DEFRA) National Farm Survey data.

The ABS collects annual turnover and purchases by businesses and estimates industry aggregates. These estimates are adjusted to meet the national accounts concept of output and intermediate consumption. This includes exhaustiveness adjustments such as for unregistered traders and the informal economy.

Industry totals for output and intermediate consumption derived from the ABS, are allocated to products using proportional patterns. For allocation of intermediate consumption, we use data from our Annual Purchases Survey (APS).

For output of the manufacturing industries, we use our UK Manufacturers' Sales by Product (PRODCOM) Survey. This is a well-established survey that gives good quality data about the products manufactured by UK businesses.

For the service industries, we previously relied on proportions that were based on ad-hoc work done in the early 2000s and updated with data from the previous Services Turnover Survey (STS). In recent years, these proportions have been fixed and have become increasingly outdated. For many industries, the product patterns were largely diagonal; that is, industries were assumed to mainly create products directly linked to their own industry, with limited secondary activity.

For both manufacturing and service industries, a range of fixed adjustments are applied to reallocate the survey-based product output. This is to cover output that is not well represented in the survey data. Specifically, output of business services, rental, and operation of workplace canteens and restaurants.

Additionally, the output of the research and development (R&D) product in the R&D industry is based on data from the Business Enterprise Research and Development Survey (BERD) with appropriate reallocation of other products in that industry. The allocation of the R&D product is not changed by work to incorporate the ASGS data since the BERD Survey remains the best source. We implemented the ASGS to capture secondary production of the R&D industry in the 2025 national accounts data release.

Allocating private non-financial corporations market output to products using the new method

The Annual Survey of Goods and Services (ASGS) data are now fully utilised to allocate industry level output, derived primarily from the Annual Business Survey (ABS), to individual products. This is an important quality improvement, although some product allocations were already partially aligned to early iterations of the ASGS. This happened during the implementation of double deflation and in the 2025 UK National Accounts data release.

The previous fixed proportions are replaced with dynamic product allocations from the ASGS. Furthermore, additional adjustments for various business services, rental, and for operation of workplace restaurants and canteens are no longer required in the service industries. This is because that output is now picked up by the ASGS. Those adjustments are still used in the manufacturing industries because UK Manufacturers' Sales by Product Survey (PRODCOM) does not cover output of service products at the level of detail required. This is likely to be subject to further development under plans to modernise our business surveys.

Benefits of the new approach include:

• business data capture both primary and secondary activities, which produces a wider and more realistic range of industry-product combinations, reducing one‑to‑one (diagonal) mappings

• it replaces historic proportions as activity is directly measured

• it maintains industry totals (the ABS remains the preeminent source with a much larger sample size) but substantially changes product allocations

• in some cases, new data explains previous product supply and use imbalance puzzles in the SU tables

About the Annual Survey of Goods and Services

The ASGS measures turnover generated by UK businesses in the service sector. This is broken down into the different goods and services from which it is generated, both inside the UK and outside the UK.

Businesses are selected (by 4-digit industry) from a directory of businesses in the UK, which are registered for Value Added Tax (VAT) or PAYE with Companies House. Approximately 40,000 businesses are sampled every year from a population of approximately 1.9 million.

The ASGS is conducted under the Statistics of Trade Act 1947. This means that, once selected, businesses are legally obliged to return data. All information provided is strictly confidential.

Businesses of all sizes are selected to ensure our statistics are representative of the UK economy. We select all large businesses (those with 300 or more employees). We also select some small and medium businesses, usually for a set period of time, then replace them with other businesses from their industry. They are very important since their trading patterns are often different to the large businesses.

The survey sample is balanced to ensure effective representation of each industry in the service sector. Industries covered (based on SIC2007) range from D351 (electricity production) through to S9609 (other personal services nec). Financial and construction industries are excluded, as is the public sector and NPISH sector.

This survey launched as a pilot in 2017 collecting data for the year 2016. Data are published in our Annual Survey of Goods and Services bulletin. Read further technical information in our Annual Survey of Goods and Services technical report.

Adjustments needed to the Annual Survey of Goods and Services for use in UK National Accounts

The 2016 pilot survey data collected during 2017 was not considered of sufficient quality to be used at the granular product level required in the national accounts. Important improvements to the questionnaire design were made at the end of the pilot - they were implemented in 2018 for collection of the 2017 survey data. This improved the quality of the disaggregated product totals required for national accounts processing.

Within the parameters of use we have outlined, the ASGS data have several quality and methodological adjustments to enable incorporation in the accounts.

Quality assurance and adjustments

Data cleaning was used to remove implausible or misreported industry and product combinations and one‑off anomalies. This was accomplished via microdata assessment either direct with businesses or via ancillary sources.

Products that were routinely misclassified by businesses were reallocated back to their appropriate categories. The public administration product, which can only be produced by government, was often entered by businesses in the legal services industry because of confusion over funding through legal aid. Specialist financial products were sometimes entered by non-finance businesses, which were more appropriate to referral fees. In both cases, the values were allocated back to main (diagonal) output. Survey processing arrangements have since been improved to help avoid these issues.

Physical goods produced by service industries were incorporated; they were mapped to the appropriate product categories using survey responses and supporting evidence.

Final enhanced quality assurance was carried out during supply use balancing by comparison with existing patterns, and investigation of large or volatile changes.

Methodological adjustments

Margin activity (wholesale and retail) turnover in the ASGS is collected on a gross basis. However, for national accounts output only the margin earned is required. An adjustment is applied to the ASGS data to remove the estimated cost of the goods sold. This adjustment is based on industry data from the ABS which already collects cost of sales data for margin activity. From 2026, the ASGS will start to collect the cost of sales information directly. This will allow a review of current margin rates and potential refinement of the method in the future.

The ASGS data are only available for use in national accounts from 2017. This left several years where fixed proportions were used for product allocation. To address the remaining fixed proportions, and to avoid a discontinuity in the product patterns, the ASGS based patterns were back-cast to 2007.

The back-cast in the unbalanced data were based on a straight-line method joining the average of 2004 to 2006 proportions to the average 2017 to 2019 proportions. This had the benefit of minimising the impact of atypical values in 2017 or 2006 on the back-cast. The choice of 2007 as the start point for back-cast was based on dynamic proportions before that point that were linked to ad-hoc survey work in the past. There was also a desire not to rewrite history based on limited new information.

Exceptions

The ASGS data do not cover the public sector or the private non-financial corporations (PNFC) sector in the following industry groupings:

• manufacturing
• agriculture
• construction
• the financial sector

Sources for product detail in those industries are therefore unchanged.

The ASGS data are available for the PNFC sector’s energy and waste management industries. However, the granular product level data were highly variable and were deemed to not be of sufficient quality to use at this point. The energy industries are already subject to additional scrutiny through collaboration with Department for Energy and Net Zero (DESNZ).

Ancillary impact of the Annual Survey of Goods and Services data

The direct impact of the ASGS data is on the product patterns for PNFC sector output estimates. However, product patterns for the public corporations (PC) sector are inferred from PNFCs, since no alternative source is available. As with the PNFC sector, industry level totals are fixed from other sources with only the product breakdowns impacted. The effect of the balancing process on this sector is much less pronounced than for the PNFC sector. This means that industry gross value added (GVA) is largely unaffected. However, the product impacts still needed to be taken in to account for the overall balancing process.

Timing of implementation

The survey data are used between 2017 and 2024 and will be back-cast to 2007. The lead time to fully integrate the ASGS data has been necessary to ensure an appropriate level of quality assurance for survey data and national accounts processes. This has been especially important in integrating survey data for the coronavirus pandemic period (2020 and 2021).

Integration was originally scheduled for implementation in our 2025 annual data release. However, we decided to delay its full implementation until our 2026 data release to allow additional time for a full quality review of balancing assumptions. This included a review of sources of data on the economic demand for products where the ASGS data suggested an important change to the level of supply in the economy. Therefore, our 2025 data release only included the ASGS's effects on some targeted cases, for example, in the research and development (R&D) industry. Our 2026 data release includes the full range of the ASGS data, following the substantial and detailed quality assurance process.

Balancing of current price GDP

As we have already outlined, each of the three approaches for calculating gross domestic product (GDP) is separately compiled before the single estimate of the current price level of GDP is derived, using the annual supply and use balancing framework.

The ASGS data had a significant effect on our assessment of product supply from the PNFC sector, and to a more limited extent, from the PC sector. Total supply of individual products was substantially altered in all periods from 2007 onwards. Rebalancing of the supply and use matrix was conducted over an extended period, and included additional quality assurance of the ASGS-based product data, using direct contact with businesses and assessment of ancillary data.

On the supply side, during balancing we considered the new ASGS data at product level and, in some cases, adjusted industry level totals for output in addition to the product breakdowns. Similarly, imports of services were reassessed in some cases in response to revised domestic output of service products. 

On the use side, existing balanced estimates were reviewed in the light of the new ASGS product data for:

• exports of services

• Gross Fixed Capital Formation (GFCF)

• Household Final Consumption Expenditure (HHFCE)

Additionally, intermediate consumption estimates were reviewed in terms of product mix and industry totals.

For production GDP, this means that product levels in production and intermediate consumption have all been reassessed, with an associated change to industry level GVA in many cases. To maintain the balance of GDP, income estimates at industry level have also been reviewed, mostly via gross operating surplus or compensation of employees. GDP expenditure changes in line with the other measures via the rebalancing of GFCF, HHFCE, and Trade (mostly services). 

The net effect of the rebalancing was a change to the overall aggregate balanced measure of GDP in current prices, by relatively small amounts in most years. GVA at industry level is changed to a larger degree but is somewhat offsetting, as would be expected.

Outline of the effect on the estimate of GDP

As we have already noted, implementation of the ASGS does not change our starting estimate (before supply use balancing) of the total output of industries in the PNFC sector as this comes from the ABS. However, the improved understanding of the range of products that service industries produced led to new patterns of product output in those industries, and different balancing decisions, impacting product patterns across multiple industries (not just service industries).

Figure 1: Heatmap showing the change in average proportional product market output within industry, UK, 2017 to 2023

Figure 1 shows the change in percentage point terms between pre- and post-ASGS average proportion of products for market output in each industry in the PNFC sector. This relates to post balancing current price data in the period 2017 to 2023 (the period for which we have data based on pre- and post-ASGS methods). It is shown in the 114 industries by 114 products matrix used for supply use balancing. Within Figure 1, the pre- and post-ASGS, average proportions are shown for every industry product combination.

The most important effect is in the service product proportions in service industries.

Diagonal industry and product proportions are reduced in favour of a more diverse range of products. There is a tendency for output to spread into related products around the diagonal, but this is not a universal trend.

The effect of removing the fixed proportion adjustment for canteens is visible as reduced proportions for multiple industries in the line for the food and beverage serving product. The ASGS has not identified replacement output for this product. This is most likely a function of workplace canteens declining in importance since the early 2000s, being replaced by self-service options (vending machines and microwaves) or by outsourced canteen operators (separate businesses in the food and beverage serving industry).

The effect of removing the fixed proportion adjustment for rental is visible as reduced proportions for multiple industries, in the line for the other real estate services product. The ASGS has not identified similar levels of replacement output for this product.

Some output of manufactured products made by service sector businesses have been uncovered through the ASGS data. While the primary focus of the ASGS is service industries and products, data are collected on all products made by service industries. This is the case especially through industrial services, which are classified across multiple products within the manufacturing segment of the classification of products by activity (CPA), despite being more broadly a service activity.

The ASGS data (extensively quality assured) also detected new and important off-diagonal output of agricultural, energy, waste management, and construction products, in addition to manufacturing products, and industrial services.

Because of this previously undetected output, there is visible change in the non-service product proportions for the service industries. To ensure the supply and use of these products remain balanced, output from the non-service industries has been adjusted in some cases.

This is shown in Figure 1 as change in the quadrant for non-service product proportions in the non-service industries. Generally, this change reduces output around the diagonal intersection of industries and products (industry output of its principal product).

Change in the product proportions post-balancing, will have some effect on aggregate current price (nominal) GDP. This is however expected to be relatively small.

Additional impact on real GDP (chain volume measures) is also likely to be observed. This is because annual double deflation methods are based on products rather than industries. The new mix of products in industries and their effect on product totals will therefore affect the outcome of deflation methodology (although the methodology itself is not changed).

Actual impacts on nominal and real GDP will be discussed as part of later articles covering all changes implemented in the 2026 Blue Book.