The Annual Business Inquiry (ABI) estimates cover all UK businesses registered for Value Added Tax (VAT) and/or Pay As you Earn (PAYE), classified to the 1992 or 2003 Standard Industrial Classification or UK Standard Industrial Classification 2007 headings listed in the tables. The ABI obtains details on these businesses from the ONS Inter Departmental Business Register (IDBR), (for more details see the heading on Register below).
Statistical and reporting units
The inquiry was addressed to the individual legal unit, i.e. the individual company, partnership, sole proprietorship, etc. In a few cases, combined returns were accepted covering two or more legal units which were not able to provide separate returns.
Forms
As with all its statistical inquiries, the ONS is concerned to minimise the form-filling burden of the ABI. The forms are designed to ensure the burden on the individual contributor has been kept as low as possible.
A number of different form-types are used in the inquiry:
Long form-types were sent to all businesses with an employment of 250 or more and also to a proportion of selected businesses with lower employment.
Short form-types were sent to the remaining selected businesses.
From 2002 Northern Ireland data is collected by Department of Enterprise, Trade and Investment in Northern Ireland (DETI) using their own forms
The forms differ in that long form-types ask for a detailed breakdown of purchases; employment costs; taxes, duties and levies, whereas short form-types just ask for the totals of these variables.
Response
Details of the response rates for all years can be found on the Quality Page
The Statistics of Trade Act 1947 prohibits the disclosure of any information relating to an individual undertaking without the consent of the person carrying on that undertaking. Rigorous checks have been made to ensure that information relating to individual businesses has not been disclosed, either directly or by deduction, in any of the figures released. Similar precautions will be taken in safeguarding the confidentiality of traders' returns in the preparation of any special analyses and in all other statistics which are made available at any time.
Rounding of figures
Where necessary, figures have been rounded to the nearest final digit shown. Consequently, there may be slight discrepancies within a table between totals and the sums of constituent items.
Methods Used
The development of the ABI is described in an article published in the November 2000 edition of Economic Trends, which is available here . This article must be read in conjunction with the methodology information given on this page.
The register
The register used for the ABI is the IDBR, which consists of companies, partnerships, sole proprietorships, public authorities, central government departments, local authorities and non-profit making bodies. The main administrative sources of the IDBR are HM Revenue and Customs (HMRC) for VAT and PAYE details.
The following can be passed to the ONS in respect of each VAT registration and for members of group registrations for use in creating a statistical register or conducting a statistical inquiry: - VAT number, name, address, trading style, legal status, standard industrial classification and turnover.
Each legal unit is classified to a single activity, whether it is wholly or mainly engaged in that activity. The nature of a business can change with time, possibly because another business has been absorbed in a take-over. Some businesses have a very significant secondary activity, perhaps completely different from their main activity, and a small change of direction can lead to a new main activity and to the reclassification of the business. Other changes may arise from improvements to classification data held on the IDBR as a result of new information received about individual businesses. The reclassification of businesses for reasons such as these contributes to the year on year changes in the figures published.
Sample Design
The sample was designed as a stratified random sample of about 66,971 businesses (2008 inquiry) from the register of legal units. The inquiry population is stratified by SIC(92)/ SIC(2003)/ SIC(2007), employment, and country using the information from the register. The sampling scheme is designed to give best estimates of the population totals for a given sample size and involves selecting all the largest businesses with a progressively reducing fraction of smaller businesses. This method ensures the sample size is kept to a minimum.
The inquiry results are grossed up to the register population, so that they relate to all active UK businesses on the IDBR for the sectors covered. For 2002 data published on 29th June 2004 and 2003 data published on 16 December 2004 Northern Ireland was collected and processed by DETINI. However 2002 and 2003 revised data published on 23 June 2005 was collected by DETINI but processed by ONS.
Estimation
Factors were produced to enable estimates for all businesses classified to each SIC(92)/ SIC(2003)/ SIC(2007) to be compiled from data provided by responding businesses. These factors were calculated for each employment size-band within each SIC(92)/ SIC(2003)/ SIC(2007) and are equivalent to the ratio of responding businesses to the total number of businesses. Northern Ireland and Scotland are sampled and estimated for separately, England and Wales are sampled separately but are combined for the estimation procedure.
Returns for the few large non-responders were estimated for individually. This estimation was normally based on summary data received from the business, or on the business's return to the inquiry in the previous year adjusted to take account of the likely change in the value of trading over the period.
Regional Estimation
Important Information about Regional Data - The total regional estimate for a variable summed across all regions within any industry will not necessarily equal the UK industry estimate of the variable. This is due to the method of calculating estimated regional data. The regional estimates are constrained to the UK data at the all-industry level.
The ABI regional methodology apportions returned data to the local units (individual sites) of the reporting unit (main site from which data is collected) before the estimation process. The estimation process is then performed at local unit level. For the UK industry level data the estimation process is performed at the reporting unit level.
The local unit estimation process calculates an estimation factor for non-response from the apportioned local unit data. The estimation factor is then applied to the apportioned local unit values to give a total estimate for the local unit population. The estimation factor is calculated using the IDBR employment, classification and region of the local unit. This estimation factor will cause the regional data to differ from the estimated national data and so the data is constrained to the UK all industry level.
The reporting unit estimation process calculates the estimation factor for non-response from the returned reporting unit data. The estimation factor is then applied to the returned reporting unit data to give a total estimate for the reporting unit population. The estimation factor is calculated using the IDBR employment, registered turnover and classification of the reporting unit.
The local unit classification may differ from the classification of the reporting unit and the apportioned value for the local unit will be added to the industry of the local unit, not the industry of the reporting unit as would be the case for UK data.
The sub-national estimates are constrained to the UK data at the all-industry level and so the total regional estimate, for a variable summed across all regions, and all industries, will equal the UK estimate of the variable at the A-O level in the national results.
Regional Capital Expenditure
For the reference year 2005 onwards, figures on regional capital expenditure will no longer be made available on the ABI website pages. Accurate estimation of regional capital expenditure is dependent on there being a strong correlation between the variable on which any estimation is based, in this case local employment, and the variable for which we are attempting to produce estimates, here regional capital expenditure. It has been established that the relationship between regional employment and regional capital expenditure is, in fact, unreliable.
Regional capital expenditure figures for earlier years will continue to be available on the ABI website, although the concerns identified apply equally to those earlier figures. ONS does not recommend the use of these regional capital expenditure figures but recognises that some users will, nevertheless, wish to have them.
It will still be possible to obtain regional capital expenditure figures for 2005 onwards by requesting either standard extracts, for which there is no charge, or special analyses, for which a charge will be made.
For more information about either of these services, please either email abi2@ons.gov.uk or telephone +44(0) 1633 456592 for standard extracts or +44(0) 1633 456601 for special analyses.
Comments on this are welcome and should be emailed to ABI2@ons.gov.uk.
Further information on the regional methodology is contained in an article published in the November 2000 edition of Economic Trends. Refinements to the methodology have taken place since this article was written, methodogical information given on these pages supersedes the Economic Trends article.
Estimates of Gross Value Added shown for regional data will not be exactly the same as those published in the United Kingdom Regional Accounts. They differ in three key areas; adjustments for coverage; adjustments needed to move the accounts onto an ESA 95 basis; and adjustments for balancing purposes. More information can be obtained here
Sampling and other errors
Estimates derived from samples invariably produce results which differ from those that would have been obtained from a complete survey. If a number of different samples were selected then each would produce a different result. Sampling errors measure the extent to which these estimates can be expected to differ from the 'true' value.
These sampling errors are small for the aggregates of the main ABI variables and indeed the sample is specifically designed to achieve this. Quality measures are available for the main variables.
However, in addition to sampling errors there is the potential for non-statistical errors which cannot be easily quantified. Examples where these errors may occur are deficiencies in the inquiry register and errors made by respondents in completing the inquiry forms.
Expansion of information on the forms
Short form-types ask only for total purchases; total employment costs; total taxes, duties and levies rather than the detailed breakdown asked for on long form-types.
Proportionate breakdowns of the full purchases; employment costs; taxes, duties and levies provided by the smallest businesses completing the long form-types were calculated for each SIC(92)/ SIC(2003)/ SIC(2007). Factors were then applied to the data against the summary headings on the small form-types to derive the breakdown for the above variables.
Discontinuities
2008
The data from 2008 is aggregated on the UK Standard Industrial Classification of Economic Activities 2007. This is a major revision of the UKSIC(03), in line with the latest revision of the European Community classification system NACE Rev.2. Further information is available in the UK Standard Industrial Classification 2007.
2006
From the publication of 2006 provisional data, there has been a change in the sampling arrangements for the two parts of the Annual Business Inquiry. In previous years ABI 2 was a sub-sample of ABI 1, which collects the employment figures. For the 2006 inquiries, ABI 1 and ABI 2 have been sampled separately.
2003
From the publication of revised 2002 and 2003 data on 23 June 2005 Northern Ireland data has been processed by ONS. For the 2003 inquiry period there was a change introduced into ONS's ABI procedures to ensure Northern Ireland data was processed on a parallel basis to GB. However it was not possible for ONS to make this change within its systems for 2002 data and this may have led to some discontinuity in the Northern Ireland estimates between the two years.
In line with ONS revisions policy, data for 2002 will not now be amended. However, as a one-off exercise, ONS has in conjunction with DETI(NI) examined the growth in Gross Value Added (GVA) between 2002 and 2003. That exercise suggests that if the two ONS inquiries had been conducted on a similar basis, the Northern Ireland GVA for the industries covered by the ABI would have been estimated to have increased within the range of 7 to 9 per cent.
The data from 2003 is aggregated on the UK Standard Industrial Classification of Economic Activities 2003. This is a minor revision of the UKSIC(92), in line with the latest revision of the European Community classification system NACERev.1.1, and in respsonse to the UK user demand for more detail at the 5 digit subclass level. Further information is available in the UK Standard Industrial Classification 2003.
The spreadsheets have been annotated in blue to show amended text, numbering or SIC content.
2002
From the publication of revised 2002 and 2003 data on 23 June 2005 Northern Ireland data has been processed by ONS.
For 2002 the Department of Enterprise Trade and Investment in Northern Ireland (DETI) have conducted their own Annual Business Inquiry. This has meant there are methodological differences between the way the results for Northern Ireland and the regions within Great Britain (GB) have been calculated for the UK regional results. This means the results produced by ONS for Northern Ireland for 2002 are not consistent with those produced for 2001 when the ABI was a United Kingdom survey. Users who want to look at trends in results for Northern Ireland for 2001 to 2002 are advised to look at the Northern Ireland website.
As ONS has calculated a GB regional analysis for 2002, rather than the UK regional analysis for 2001, it means that there are also discrepancies when comparing GB regional figures for 2001 and 2002 but because Northern Ireland represents only 2 per cent of the UK economy these discrepancies are minor.
2001
From 2001 coverage of Section A was further extended to include Groups 01.4 (Agricultural and animal husbandry service activities, except veterinary activities; landscape gardening) and 0.15 (Hunting, trapping and game propagation including related service activities).
From the release of provisional 2001 data within Division 50 (Motor Trades) there is a discontinuity between 2000 and 2001 data. This is due to the reclassification of several businesses from Division 50 in 2001. Without these reclassifications it is estimated that 2001 data would be around 3-4% higher. As the businesses are being reclassified outside of the classifications covered by the ABI Financial data there is no equivalent discontinuity elsewhere in the data.
2000
From 2000 ABI coverage was extended to include Division 02 of Section A (Forestry, Logging and Related Service Activities) and Division 05 in Section B (Fishing).
1999
From 1999 improvements to the questionnaires for Sections M (Education - excludes some public sector) and N (Health and social work - excludes some public sector), have led to a movement of data from 'total turnover' to 'value of grants/donations received'. This has caused the approximate Gross Value Added at basic prices to fall between 1998 and 1999. Data from 1999 onwards has been collected and processed on a consistent basis.
1998
Scotland has been sampled and estimated for separately from England and Wales.
From 1998 data for Group 64.1 (Post and courier activities) includes data for 64.11 (National Post Activities), this activity heading was excluded from ABI prior to 1998.
1997
From 1997 in Group 63.3 (Activities of travel agencies and tour operators; tourist assistance activities not elsewhere classified) total turnover, total purchases and approximate gross value added at basic prices includes monies received and paid in respect of tour operators' invoices. This differs from earlier years where only sales on own account transactions and the commission from tour operators were recorded.
1995 and 1996
Data were calculated on a different basis from those for 1997 and later years. In order to provide a link between the two data series, the 1995 and 1996 data have been reworked to provide estimates on a consistent basis. The main changes are the treatment of outliers and the level of expansion have been amended since the 1996 inquiry.
Symbols and abbreviations
The following symbols and abbreviations are used within the ABI data:
Introduction to ABI
