Voluntary sector definition
As in previous Almanacs, the “general charities” definition is used to obtain estimates for the voluntary sector. The definition is based on common features of non-profit organisations and was originally constructed to also fit Office for National Statistics (ONS) national accounting purposes.
This definition excludes registered charities that do not meet these criteria, for example sacramental religious bodies or places of worship as well as organisations like independent schools, government-controlled bodies or housing associations.
NCVO recognises the limitations of the general charities definition; a recent internal review identified the need to consider expanding the range of organisations included in the Almanac in the future, whilst also maintaining a time series for general charities. For the moment, this is addressed by including a range of other civil society organisations in the Almanac section on ‘Civil Society’.
Voluntary sector methodology
Key data source
Financial information on voluntary organisations is based on their annual accounts submitted to the Charity Commission. The Charity Commission maintains a register of charities, with details of all organisations based in England and Wales that are recognised as charitable in law. This register is therefore the ideal sampling frame for selecting samples of registered charities. All but the smallest charities are required to provide an annual return with their accounts and trustees’ annual report; a sample of those annual accounts forms the basis of the Almanac data set.
The “general charities” definition is applied to all charities on the Commission’s register, producing a usable population for England and Wales.
Financial data for a sample of just over 7,800 of these organisations was obtained for the Almanac 2017 by entering data from the charities’ annual accounts. This data entry process was carried out on behalf of NCVO and the Third Sector Research Centre by the Centre for Data Digitisation and Analysis at Queen’s University, Belfast.
The sample design was originally based on taking a random sample of general charities stratified by the size of organisation in terms of their annual income, because this variable is both a key determinant of sampling error and a key variable for analysis. Different sampling fractions are applied to the different sizes/strata, the fractions increasing with size, until for ‘major’ organisations (with incomes of more than £10m) all organisations are sampled. Data is weighted at the analysis stage to take account of the different sampling fractions.
|Registered with the Charity Commission||77,806||55,135||25,163||5,737||1,093||75||165,009|
|Sample (% of general charities)||0.1%||2.5%||14.6%||77.4%||89.7%||100%||
The aim of this design was to minimise sampling variability in the data for both totals and proportions. The decisions about the sampling fractions to be applied were very much a balance of practical, analytic and precision issues. A joint NCVO/TSRC Working Paper gives more details. The sample was designed to produce reliable figures for subgroups of the three main analysis variables: income band, region, and ICNPO classification. It was not designed to produce reliable figures at the intersections of these categories e.g. the finances of recreation and culture organisations in the East Midlands.
Subsequently the sample design was taken forward as two ‘panels’ over alternating years, except for general charities with incomes of more than £10m that are included every year. This came about because charities are required to restate the previous year’s figures in their accounts as well as providing the current year’s accounts. For some charities (those that are sampled every year, or for two consecutive years, mainly the larger ones) we are able to collect both the ’current’ and ‘previous’ years, hence forming a quasi-panel. For such charities, we collect information about a particular financial year twice; data for the 2013/14 financial year, for example, can be collected once as the ‘current’ year in the 2016 Almanac and once as the ‘previous’ year for the 2017 Almanac. We input the data for both years, and compare the ‘previous’ year in our new dataset with the ’current’ year from the old one, as part of our checking procedures. We also include the ‘previous’ year’s restated information in our aggregate data, and the impact of this is discussed further below.
Impact of modifications
Current practice is that, to provide us with a larger overall sample, we include the ‘previous’ year’s data from the latest year in the aggregate data for the previous year. For example, in the Almanac 2017, we are focussing on data for 2014/15 but where we also have restated data for an organisations for 2013/14, that is included in the aggregate data for 2013/14. We are thus ‘revising’ the values that had been reported for 2013/14 in the previous Almanac 2016. This revision is, however, not always an improvement; it can give rise to inconsistencies because we do not sample the same organisations for each income band in each Almanac so there will be differences in which restated accounts we use for some income bands. We are currently reviewing whether the benefits of the additional sample size outweigh the potential inconsistency.
Before use, the data is cleaned to remove significant errors, and undergoes a series of checks to ensure validity.
These checks include:
- Comparison of income, expenditure, assets and workforce data between this year and last year to look for particularly large increases and decreases;
- Construction of various ratios between financial variables (for example between income and expenditure, and investments and dividend income) to look for anomalies;
- Manual checking of annual accounts submitted by major and super-major charities.
Those records, where accounts were submitted in a foreign currency, were converted to Pounds Sterling using an average of the exchange rate over the year. Organisations have a range of financial year ends, distributed throughout the year. To ensure consistency, all values were converted to April 2015 prices using the retail price index (RPIX).The retail price index (RPIX) was also used for trend data to convert actual values from previous years to April 2015 price.
Once the data is cleaned, mean amounts are produced for all financial variables within each income band and are multiplied to the England and Wales population size using weights based on income bands.
Supplementary data from SCVO (Scottish Council for Voluntary Organisations) and NICVA (Northern Ireland Council for Voluntary Action) is used to produce estimates of the UK population. Due to rounding figures, some percentage totals may not sum to 100%.
Analysis by sub-sector
Sub-sectoral analysis is based on assigning charities to categories in the International Classification of Non-profit Organisations (ICNPO).
The International Classification of Non-profit Organisations (ICNPO) is a classification system for non-profit organisations and was designed by the Center for Civil Society Studies at Johns Hopkins University in US as part of efforts to draw up a UN Satellite Account for the non-profit sector. It is the most useful tool to classify and compare different groups of voluntary organisations, and is used within the Almanac.
The classification involves a variety of different methods including keyword searches, matching to other registers and looking at individual sources. As the original system was not a perfect fit for the UK voluntary sector, a few categories for large groups of organisations (such as scout groups and nurseries) were added. For the figures presented in this Almanac some categories have been grouped together.
In reality, many organisations undertake multiple activities (e.g. housing and advice), but the ICNPO groups organisations into a single category based upon their primary activity. The advantage of the ICNPO over a multi-dimensional classifications system (such as the classification system used by the Charity Commission), is the ability to look at and compare discrete groups of charities. The classification is not perfect. It is a classification of charities rather than claiming to be the classification. However, it does allow for the comparison of groups of charities and it does cover the whole sector.
Analysis by income band
Within the Almanac, voluntary organisations are divided into six groups based on their income. Each group is named to make it easier to discuss the findings and place them in context. The sample data, however, is gathered in nine bands to accommodate Charity Commission registration thresholds. These bands are aggregated to produce the six bands used in the Almanac.
In 2016, we introduced a “super-major” group including charities with more than £100m annual income. This income band was warranted because there had been a noticeable increase in organisations with income of over £100m.
|Income||Income (sample bands)||Category|
|Less than £10,000||Zero income||Micro|
|£1 - £10,000|
|£10,000 to £100,000||£10,001 - £25,000||Small|
|£25,001 - £100,000|
|£100,000 to £1 million||£100,001 - £500,000||Medium|
|£500,001 - £1,000,000|
|£1 million to £10 million||£1,000,001 - £10,000,000||Large|
|More than £10 million||£10,000,001 - £100,000,000||Major|
|More than £100 million||Over £100,000,000||Super-major|
Other data sources
Charitable giving data is from the CAF UK Giving 2015, based on a survey of 4,160 individuals conducted by GfK NOP.
Our employment figures are largely based on Labour Force Survey (LFS) data, which is the only national data source that attempts to classify individual employment by sector. The LFS surveys an estimated 60,000 private households every quarter. By pooling data for unique individuals from four quarters, it is possible to produce reliable estimates of the voluntary sector’s workforce. Weighting is used within the LFS to compensate for non-response rates in certain groups and produce population estimates.
To identify the sector a respondent is employed in, a two-stage self-classification process is used. Respondents are first asked whether they work for ‘a private firm, business or a limited company’ or ‘some other kind of organisation’. Those respondents who choose the second option are then asked, ‘what kind of non-private organisation is it?’. They are then presented with a range of options including ‘charity, voluntary organisation or trust’. For the purposes of the analysis for the Almanac, responses to these questions were recoded into a sector variable and defined as ‘private’, ‘public’ or ‘voluntary’.
This data draws on the Citizenship Survey (2001-2010/11) and Community Life Survey (2012/13-present), the best sources of data on rates of volunteering in England. There was a gap in the data when there was no survey in 2011/12, during the transition between the two surveys. The Community Life survey is commissioned annually by the Cabinet Office and carried out by TNS BMRB, and is designed to be representative of adults aged 16 and over in England. The measures used here were common to both surveys. The data collection methods were also broadly similar.
The survey describes both ‘formal volunteering’, which takes place through a group, club or organisation, and ‘informal volunteering’, which takes place independently of such structures. Data is drawn from the most recent survey unless otherwise stated, which reports on volunteering during the 2015/16 year, and includes the appropriate weighting.
There are many types of organisations which are not registered charities but which inhabit the space between the state, businesses and individuals. This is often referred to as ‘civil society’. Civil society includes large groups of specific organisations, such as universities, trades unions and housing associations, which are clearly different from charities and other groups where the boundary with charities is somewhat blurred, such as faith groups, benevolent societies and companies limited by guarantee.
This section draws on many different data sources to produce figures for
- Benevolent societies
- Building societies
- Common Investment funds
- Community Interest Companies (CICs)
- Credit Unions
- Employee Owned Business
- Football/Rugby supporter trusts
- Friendly societies and mutual insurers
- Housing Associations
- Independent Schools
- Leisure Trusts
- Political parties
- Religious bodies
- (Social) Companies Limited by Guarantee (CLG)
- Sport Clubs
- Trade associations and professional bodies
- Trade unions
- Unincorporated associations
Please refer to the Civil Society methodology for more detail.
- Salamon, L. M., Anheier, H. K., List, R., Toepler, S., & Sokolowski, S. W. and Associates (1999) Global Civil Society: Dimensions of the Nonprofit Sector. Baltimore, MD: Johns Hopkins Center for Civil Society Studies. / Kendall, J., & Knapp, M. (1996) The Voluntary Sector in the UK. Manchester: Manchester University Press, pp.316.
- The 1994/1995 ONS survey of Charitable Organisations was carried out in partnership with NCVO for the purpose of preparing data about the voluntary sector for National Accounts.
- More specific options for expansion will be considered over the next 12-18 months
- ‘Based in England and Wales’ is defined as either holding most of their assets in England and/or Wales, or having all/the majority of trustees normally resident in England and/or Wales, or being companies incorporated in England or Wales
- For 2014/15, the requirement to file annual accounts and the trustees’ annual report applied to registered charities with an annual income over £25,000. Those with annual income of £10,000 – £25,000 are required to provide only an annual return and those with an income of £10,000 or less are asked to complete the annual return for certain items only. Thus although the smallest charities are registered with the Charity Commission, information about their annual accounts is not normally available.
- Nine income bands are used: zero income ; under £10K ; between £10K and £25K [2.1]; between £25K and £100K [2.2]; between £100K and £500K [3.1]; between £500K and £1m [3.2]; between £1m and £10m ; between £10m and £100m ; over £100m . See also Analysis by sub-sectors.
- In principle, the sampling fraction applied is 1 which means that all are included. In practice a small number may sometimes be excluded because data is not available (if, for example, they have failed to return accounts to the Charity Commission).
- The tendency for different samples, even if they are random, to give slightly different answers.
- TSRC Working Paper 93 http://www.birmingham.ac.uk/generic/tsrc/documents/tsrc/working-papers/working-paper-93.pdf
- This allows people reading the accounts to get an idea of how things have changed since the last year.
- For example, for the data published in the 2017 Almanac we collect both data for 2014/15 (‘current’) and 2013/14 (‘previous’) years.
- The difference between £5.39bn in the Almanac 2016 and £6.08bn in the Almanac 2017 (both at 2014/15 prices)
- The difference can be exacerbated by the fact that a relatively small proportion of smaller charities are sampled (for example, 15% of medium sized charities are sampled) so their accounts, including levels of grant spending, are weighted up in the overall sample. It is therefore possible for a small number of different values (outliers) to have a disproportionate impact. For example, one particular organisation had a grant payment of £11m (for which there was no equivalent in the 2013/14 accounts) and this was weighted up to almost £100m for the whole population.