TSRC Research Digest:
Building the research infrastructure
Old SPSS files at NCVO, Jaz drives, and our panel data on registered charities
John Mohan writes:
A little-known aspect of TSRC’s work, but something which we believe is very important to the research infrastructure, concerns the construction of data resources on the sector. Although much data is gathered in a fairly routine fashion on voluntary organisations, for example by the Charity Commission, or through Companies House, not all this is easily available, and no individual organisation has been responsible for preserving data. Yet without reliable information on the numbers and types of organisations in the sector, our ability either to describe patterns in it, and more importantly, to analyse change over time, is limited.
An early piece of work we did involved assembling data from old versions of the register of charities for England and Wales. The Charity Commission began to computerise its register in the early 1990s, and at various dates thereafter supplied copies of it to the NCVO research team. We discovered old SPSS files going back to the mid-1990s which had been archived at NCVO. This provided headline income and expenditure on several thousand charities from around 1995 onwards. Some day, when we get the chance to explore them, there are other files which appear to be responses to questionnaires to the Central Statistical Office study of the sector in the early 90s, so we hope to be able to look at how the organisations featured therein have developed over time.
Subsequently, Les Hems, who had been at NCVO around 20 years ago, discovered some old mini tapes (Jaz drives); these were no longer readable on modern computers, but a retired computer scientist in Southampton put us in touch with someone with a collection of old hardware including a tape drive from which we were able to download older versions of similar files. This bit of work was partly funded through a related ESRC grant to John Mohan through the Centre for Charitable Giving and Philanthropy. Cleaned up these proved to contain records dating from 1992, 1993 and 1994. This isn’t exactly the nonprofit sector’s research equivalent of the Rosetta Stone, but it does take us probably as far back as we are going to go in terms of capturing a time series of data on charities in England and Wales (although some data does exist for smaller subsets of organisations, such as international development charities).
We then set to work on checking for overlaps and duplicate entries in the data, spending time on such minutiae as whether a zero meant no income at all or that a charity had just not reported anything for any given year. This was a laborious process indeed, generating a log of actions taken to clean the data which runs to several hundred pages in length; if anyone is interested we are very happy to supply it! What we have got as a result is a panel of observations of registered charities which contains between 90,000 and 135,000 organisations in any one year for the period between 1995 to the present day (where one observation means a nonzero financial return for that year), and between 12,000 and 65,000 returns for 1991 – 4. Analysts can use this to track many thousands of organisations continuously over time. For example, 27,000 charities appear continuously in the data from 1995 - 2011 and this makes serious long-term analysis eminently possible. Now that major funders like BIG have released their grants data, one could look at how the availability of funding related to the emerging financial landscape of charities.
A very good example of what is possible is work on what became known as the Tescoisation debates, prompted by comments from Ian Duncan Smith about en emerging concentration of resources among a small number of large charities, at the expense of small and local entities. By tracing a panel of thousands of organisations from 1997 – 2008, we offered a very different perspective on this. It was certainly true that much resources were concentrated in the larger sections of the charitable population, but that is probably always been the case. The question really posed was whether or not the large organisations were growing at the expense of small ones. David Clifford and Peter Backus showed, in a paper subsequently published in the Journal of the Royal Statistical Society, that while it appeared large organisations were growing more rapidly, it was questionable whether this was genuinely at the expense of small organisations, since the latter were still growing in real terms, albeit not as rapidly as the larger organisations.
We have also used this database to generate economic indicators about the state of the voluntary sector – for example, we used it to demonstrate recessionary impacts on the sector, with a rather basic indicator (the proportion of charities experiencing an increase in their income from one year to the next). It’s not a great indicator – one obvious criticism is that an extra pound of income regardless of the size of an organisation might mean that income does go up comparing one year to the next - but as a broad brush description of the pattern of change over time it was quite useful for illustrative purposes. What we showed was that in the period of relatively rapid financial growth for the sector during the economic boom years of the late 90s and early 2000s, a clear majority (often over 55%) of charitable organisations experienced financial growth in any given year. However, the proportion experiencing any growth around the time of recession dropped quite substantially. It has still to recover.
A dataset like this takes time to assemble, and because it relies on a supply of data from the charity commission, there is a time lag – so at the time of writing (December 2013) the most recent data that we have cleaned up and added to our database relates to 2011-12 financial year. On the other hand it is robust financial data, based on regulatory returns, so is as valid and reliable an indicator of the relative fortunes of charities as we are going to get. Now that we’ve made the initial effort to construct this database, it’s a relatively straightforward task to continue to add to it. With 2 million observations already, and over 130,000 being added each financial year, it’s a major resource which we hope to continue to develop over time. We are obliged to share it as a condition of our funding, and it will be deposited eventually at the UK data service.