In the late 1990s, the UK’s Department for Culture, Media and Sport (DCMS) famously introduced the idea that 13 creative sub-sectors as wide ranging as the performing arts, film, advertising and software could be grouped together and labelled ‘creative industries’.
These sectors were described as “those industries which have their origin in individual creativity, skill and talent and which have the potential for wealth and job creation through the generation and exploitation of intellectual property.”
Soon after, the DCMS began publishing regular economic estimates to support this idea, which matched, wherever possible, the creative sub-sectors to the official Standard Industrial Classification (SIC) codes used in the construction of the national accounts.
The estimates also included an allied set of occupational codes drawn from the official Standard Occupational Classification (SOC) which were also labelled as ‘creative’. In other words, a set of industries and occupations were deemed ‘creative’ for measurement purposes and, by implication, the remaining industries and occupations were not.
This development is widely regarded as having been successful in raising the economic profile of the creative industries. In the eyes of policymakers – judging by how rapidly other governments adopted the DCMS metrics – and in the eyes of others, like educators – where we have seen an explosion in university courses claiming to develop talent for creative industries.
However, the non-systematic nature of the DCMS’s earlier procedures created a large number of problems too.
Some of these were technical in nature. For example, the DCMS statisticians applied ‘weights’ to some sectors to recognize that not all of their activity was in fact ‘creative’. But the validity of those weights was not clear – including to the national statistics authorities. It was also unclear what consistency there was between the selection of occupations deemed ‘creative’ and of sectors deemed ‘creative’.
But the DCMS’s approach had conceptual, deeper problems too:
Firstly, theoretical ambiguities around key terms such as ‘creativity’ and ‘intellectual property’ led to inevitable debates about why certain sectors were included, and others were not.
Secondly, a reluctance on the part of the DCMS to publish separate economic statistics on the ‘cultural’ industries – despite having the word ‘culture’ in its Departmental title – contributed to the conflation of the concepts of ‘creative’ and the ‘cultural’.
Thirdly, and most fundamentally, the lack of a transparent process for identifying which industries and occupations should be classified as ‘creative’, and which should not, meant that the classifications went out of date.
More and more interests saw problems in the statistics – whether these were areas like Crafts and Design with a disproportionate number of under-represented freelancers working across multiple sectors including those not traditionally viewed as ‘creative’; or new and emerging industries like Video Games which were not well captured in the SIC. The absence of explicit classification criteria made it impossible to translate the measurement concerns of sectors such as these into action by the DCMS.
The lack a transparent method for classifying creative industries and occupations also held back the development of any international standard, resulting in a plethora of different classifications in different countries.
For all of these reasons, Definition, Classification and Measurement of creative industries and occupations has become a key focus of Nesta’s research effort in recent years. We’ve developed a framework – the Dynamic Mapping – which addresses the main weaknesses in the DCMS’s earlier work.
The approach is made up of three steps:
In step 1, explicit judgements are made on which occupations in the workforce should be treated as creative. We define creative roles as those which deploy cognitive skills to bring about novelty whose final form cannot be fully specified in advance. In the original Dynamic Mapping report we based these judgments on a subjective scoring of each standard occupation code in the UK workforce according to a handful of intuitive criteria derived from our reading of the different literatures on creativity. In a more recent paper, we have used detailed data on job task descriptions and machine learning techniques to label (a wider set of) occupations as creative.
Armed with this list of creative occupations, in step 2 we compute the % of the workforce that is in a creative occupation for all industries in the UK economy (in other words, its ‘creative intensity’).
And in step 3, we analyse how this creative intensity is distributed across different sectors, and on this basis partition industries into ‘creative’ and others. Specifically, we label as creative industries those with exceptionally high creative intensities. We then define employment in the ‘creative economy’ as employment in the creative industries plus those working in creative jobs in sectors outside of the creative industries.
It turns out that there are a relatively small number of industries in the UK sharing the common characteristic of employing proportionately very large numbers of individuals in creative occupations, with creative intensities of 30, 40 and in some cases 80 or even 90%. This compares with the vast majority of other industries in the UK with an average creative intensity of 3% or so. This result is important, as it suggests there is a strong statistical basis for considering ‘creative’ sub-sectors with otherwise very different cultures, business and operating models as a coherent group for policy purposes.
The DCMS consulted on and adopted the main principles of the Dynamic Mapping framework in 2014, though in a small number of cases it made different judgements on which occupations should be classified as creative and which industries should be classified as creative on the basis of their creative intensities. Dropping the use of arbitrary weights in their selection of industries also enabled the ONS to give the estimates the ‘official’ stamp.
As a result of the adoption of the approach, we can understand, for example, the geography of the UK’s creative economy workforce and where it agglomerates on a basis that is consistent with how other parts of the economy are measured. So, we’ve learned that in 2013, 43% of the UK’s creative economy workforce was employed in London and the South East of England compared with 32% of the high-tech economy workforce and just 28% of the workforce as a whole.
And we’ve been able to derive comparable international statistics, because of the use the approach makes of SIC codes, and of labour force surveys of the type that are commonly undertaken in other countries too. In coming weeks Nesta will be publishing studies of the European and North American creative economies. An important finding of those papers is that creative intensity can be used to discriminate between creative and other industries in all of the countries we have studied, which suggests that the approach may have legs as an international standard.
Another important feature of the Dynamic Mapping approach is that it acknowledges, and in principle tracks, the idea that sectors are fast changing. Specifically, the creative intensity of industries is time-varying – as industries become more or less creative, depending on how technology and other structural changes lead them to alter their workforce compositions: the mapping is ‘dynamic’.
A big advantage of the Dynamic Mapping – and the main reason why it was developed – is that it uses official data and, in using the SIC and SOC codes, it is fully consistent with the National Accounts. This is crucial, as its puts estimates of the creative economy on the same footing as estimates of other parts of the economy – which is important for policy analysis.
But the National Accounts have well-known limitations when it comes to capturing and fully representing the economic contributions of creative activity. A major concern is that the sample frames of official surveys and the nature of administrative data sources means that they do not always pick up freelancers and small creative businesses. A second problem, as mentioned earlier, is that the SIC codes are inadequate in key areas, meaning that even if the businesses are captured in the data, they may not be classified in a way that allows their contribution to be accurately identified. In such cases, a potentially fruitful area of research is to scrape online big data, such as from company websites, which is another area where Nesta is doing a great deal of research. See, for example, here and here.
Measurement work sometimes invites the question of “So what?”. Here, reflection on the evolution of new industries points to the role that Definition, Classification and Measurement can play in their legitimation – in the eyes of policymakers but suppliers, educators and investors too. In fact, it may turn out that the development of rigorous mapping frameworks is one of the primary ways in which government can support new industries – ‘mapping as innovation policy’.