Executive summary

The creative industries are one of the most important contributors to the UK economy. So it is important that we accurately measure their contribution to economic activity. Doing so can help both policymakers and industry professionals to communicate key concepts, share reliable data and make the case for greater investment.

There have been renewed attempts to estimate the true size of the creative economy. The Department for Culture Media and Sport (DCMS) and the Greater London Authority (GLA) both published studies in 2007. This report complements their work and seeks to improve on the available data about the true extent of creative activity within the economy.

What are the creative industries?

The creative industries have been hampered by multiple definitions and a lack of consistent treatment on what is classified as creative activity. The DCMS defines the creative industries as those "which have their origin in individual creativity, skill and talent and which have a potential for wealth and job creation through the generation and exploitation of intellectual property". It identifies thirteen different sectors.

The 2001 household census – the most recent census conducted in the UK – allows eleven creative industries to be identified, covering both businesses that create cultural 'products' including the arts, films and interactive games, and those providing business-to-business services in areas such as architecture, advertising, marketing and design, and web and software development. The term embraces radio and TV, news agencies and publishing, as well as jewellery manufacture and museums.

From the 2001 census data, we can identify 26 creative occupation groups. They include: town planners and graphic designers; advertising managers and furniture makers; actors and librarians; journalists and software professionals; architects and archivists. These occupation groups have evolved over the years, as some professions did not feature in the 1981 or 1991 censuses. Those working in creative jobs do not always work in creative industries. The DCMS estimates that in 2006 there were 800,000 creative jobs outside the creative industries adding to the estimated 1.1 million people working in the creative industries.

The Creative Trident

The model of a 'Creative Trident' brings together those working in the creative industries and those working in specialist creative jobs in other firms and organisations. By developing this model, we can analyse the true number of people employed in creative activities and industries and their average incomes.

We focus on three types of employment: 'specialist' artists, professionals or creative individuals working in creative industries; 'support' staff in those industries providing management, secretarial, administrative or accountancy back-up; and creative individuals 'embedded' in other industries not defined as 'creative'. Collectively, they are the 'creative workforce'.

How our model is different

CCI originally developed this model in 2005 for Australia. It differs from previous models in three key respects: it uses population data (the number of people employed in each occupation within every industry) to provide more accurate estimates; it employs a conservative approach to the selection of 'creative' occupations and industries, to avoid overreach and enable better comparability between the segments and to the economy as a whole; and it allows us, for the first time, to estimate creative incomes. However, to compensate for difficulties with the frequency and level of aggregation of UK census data, our methodology combines population data with that provided by the Labour Force Survey (LFS).

To ensure that the estimates are as robust as possible – and that we can make valid comparisons – we have focused on those who add most creative value to a process, excluding such activities as the manufacture of TV sets or musical instruments that employ small numbers of creative professionals. By doing so, we find, for example, that in 2001 only 37 per cent of those employed in publishing were in creative occupations; the same applies to 46 per cent of those in advertising or 44 per cent in radio and TV activities.

We have also excluded those occupations which are ostensibly creative but which have very low rates of employment within the creative industries. Using this sort of definition, we exclude 14 groups included in DCMS calculations – including public relations officers, ICT managers and printers – but add five new groups, including draughtspersons, software professionals and librarians.

Why this is a better model

This model offers policymakers five advantages over other models.

• First, it focuses on core creative added value, excluding activities in related chains that are not central to the creative process, such as distribution or retailing.
• Second, it enables us better to map the extent of creative individuals working in other sectors.
• Third, it distinguishes between creative individuals and others who work in creative industries, a useful tool for skills and business development.
• Fourth, it uses census data rather than sample surveys, wherever possible.
• And finally, it enables us to determine the total personal earnings arising from creative employment, a useful indicator of its economic value.

Table 1: The UK Creative Employment Trident for 2001 based on CCI analysis of ONS census data.

Table 2: Long-run growth rates of creative employment in the UK workforce 1981-2006 based on CCI analysis using census and LFS data.

Our main findings

The creative economy accounts for over 7 per cent of UK employment, consistent with the official estimates: In 2001, creative employment accounted for almost 1.9 million people or 7.1 per cent of UK employment, a very similar level to that determined by the DCMS Economic Estimates for the same period using a different methodology, definition and datasets. Table 1 shows how this is broken down within our Trident categories.

Creative employment has grown strongly over the long run: Census and LFS data together suggest that UK creative employment grew by 3.2 per cent per annum from 1981 to 2006, compared with 0.8 per cent for the broader UK economy (Table 2). The highest growth rates have been among 'specialist' creative workers within the creative industries, which have grown on average since 1981 by 6.2 per cent. However, since 2001 overall growth slowed to 1.0 per cent, just below the UK workforce annual rate of 1.2 per cent for the same period.

When we compare our estimates with those in the DCMS Economic Estimates, our estimates suggest more people working in the creative industries (in 2001, 1.2 million people versus 1.1 million) and lower levels of embedded employment (645,000 versus 787,400). Because we can show where the embedded employment occurs across the broader economy, we feel that the CCI calculations are the more robust.

When we break these data down into our creative segments, we find the largest growth has been, perhaps unsurprisingly, in the software, computer games and electronic publishing segment, which only accounted for 14 per cent of the creative workforce in 1981, but now accounts for 31 per cent, and has grown by an average 6.5 per cent a year between 1981 and 2006 (although growth from 1981-2001 was an even stronger 8.0 per cent per annum).

This segment is closely followed in growth terms by advertising and marketing, up from 5 per cent to 11 per cent of creative employment since 1981, or 6.3 per cent annual growth. Music and the performing arts has maintained a 10 per cent share of employment, with 3.5 per cent annual growth. But architecture, visual arts and design; film, TV, radio and photography; and publishing, while growing by over 1 per cent in employment terms each year, have become less important in creative employment overall.

Table 3: The mean annual income of the creative segments and modes of employment for 2006 based on CCI analysis of LFS data.

Note: Mean annual income for UK economy as a whole is £21,060

Creative incomes are higher than average

Creative incomes were on average approximately 37 per cent higher than in the UK economy as a whole in 2006. But they have grown at the slower rate of 2.5 per cent per annum since 2001, compared with 3.5 per cent for the total workforce. Creative occupations generated over £40 billion in salaries and wages in 2006, while support staff in creative industries earned an extra £16.8 billion.

Overall, the creative workforce earned 9.6 per cent of all UK earnings. Their earnings were above average, with those in software, computer games and electronic publishing employment - particularly support workers - earning most. As Table 3 shows, the earnings of embedded music and publishing workers were below the average for the UK as a whole.

More creative people work outside the creative industries than inside them

Compared with other economic activities, creative employment occurs disproportionately outside the creative industries themselves, a finding consistent with other NESTA research. Some 35 per cent of the total creative workforce (a figure that includes support staff in creative industries) is employed in non-creative sectors. This level is similar to the 39 per cent of total UK financial services workforce employed in non-financial services industries in 2001.

Such embedded creative employment is greatest in the 'manufacturing', 'real estate', 'business activities', 'wholesale and retail trade', and 'financial intermediation' industry groupings. Relative to the size of their overall workforces, outside industry groupings which include creative industries, embedded creative workers appear to be more important in financial intermediation than in any other industry grouping.

This analysis contains many useful insights. We can see how important advertising and marketing jobs are to real estate or research and development activities, for example, or the significance of creative ICT occupations to legal, accounting and business consultancy services. We can also measure employment growth or losses more accurately within each sector, and the extent of embedded creative employment from one year to the next.

Such data also provide an important lesson to policymakers: they should recognise the limitations of using industry-based approaches alone to supporting the creative economy.

Applying the methodology to financial services

The methodology used for creative industries can as readily be applied to financial services, allowing a comparison between the two sectors.

This is particularly interesting in the light of recent research which has indicated that the creative industries are as large as the financial services sector in the UK. The Financial Trident – including those working in financial professions outside the financial services sector – encompasses two million people, 150,000 more than the Creative Trident. Two thirds of the 1.2 million specialists work outside the core Financial Services industry.

By studying changes within the Trident we can see, for example, the impact of computerisation and ATMs, where clerks, tellers and book-keepers fell from 2.3 million people to 1.2 million people between 1981 and 1991. The role of financial consultants is also apparent when we break down the embedded financial workers by sector: they account for many of the almost 300,000 financial specialists employed in 'real estate, renting and business activities'.

Intriguingly, we find that sectors making use of financial expertise are also more likely to employ creative specialists, perhaps suggesting base level service requirements for efficient working in the 21st century.

Difficulties with UK data

UK data have significant limitations.

• There is no single dataset with the full range of employment and earnings information, unlike US, Canadian, New Zealand and Australian statistics.
• LFS data for smaller occupational and industry groups are less robust than census data.
• There is no census and LFS information on the earnings of the self-employed, in a sector where 28 per cent of people with creative occupations are thus employed.
• UK census data can also be dated as the information is only collected every ten years.
• There are too few industry classifications relevant to the creative industries compared with other countries, and the level of detail is often restricted, even in large occupational groups.
• The LFS is equally restrictive in its level of data, partly due to its comparatively limited sample size: approximately 350,000 individuals in 2006.

These factors lead to greater margins of error than is the case with Creative Tridents calculated in Australia and New Zealand. We need to see better data available in the future within the UK.

Recommendations for improving the data

We have been able to compare the effectiveness of the Creative Trident in both Australia and the UK, and this offers us lessons on how the UK data could be improved. The DCMS and the Office for National Statistics should work together to provide better data and improve the usefulness of their datasets.

• It should be possible to improve information within datasets by increasing the resolution – the extent to which we can drill down the data into smaller subsets – and providing finer classifications in the census 2001 data where more than 50,000 are employed in a single industry.
• The sample size of the Labour Force Survey should be increased to improve its accuracy and to allow for more detail about occupations within industries. A more frequent census of those in work including their income or a much bigger LFS should be conducted at least every five years.
• A stronger metrics culture would benefit from greater access to the source data for all official reports. Source data for the DCMS Economic Estimates and this report should be available for others to build on and challenge. Census and LFS data should be fully downloadable.

Acknowledgements

This report was written by Peter Higgs (Senior Research Fellow, CCI), Professor Stuart Cunningham (Director, CCI) and Hasan Bakhshi (Senior Policy Analyst, NESTA). The Creative Trident methodology used in this paper was developed by CCI. CCI is the Centre of Excellence for Creative Industries and Innovation at the Queensland University of Technology in Australia (www.cci.edu.au).

Additional contributions were made by Jason Potts and John Hartley (CCI) and Kate Morrison (Volterra Consulting Ltd). Ed Pickering (DCMS) and Conor Ryan made helpful suggestions.

List of tables

• Table 1: The UK Creative Employment Trident for 2001 based on CCI analysis of ONS census data.
• Table 2: Long-run growth rates of creative employment in the UK workforce 1981-2006 based on CCI analysis using census and LFS data.
• Table 3: The mean annual income of the creative segments and modes of employment for 2006 based on CCI analysis of LFS data.
• Table 4: The 11 creative industries selected from the 2001 census data.
• Table 5: Creative occupations selected from the 2001 census data.
• Table 6: The correlation between CCI segments and DCMS Economic Estimates sectors.
• Table 7: Classifications used to determine the creative core industries using census and LFS data and the percentage of employment that is within creative occupations.
• Table 8: Creative core occupations used in census and LFS datasets and the percentage of employment occurring within creative industries in 2001.
• Table 9: Creative industry-relevant classifications from the 2001 census.
• Table 10: Creative employment-relevant industry and occupation classifications from the 1981 and 1991 censuses.
• Table 11: The UK Creative Employment Trident for 1981 from the census.
• Table 12: The UK Creative Employment Trident for 1991 from the census.
• Table 13: The UK Creative Employment Trident for 2001 from the census.
• Table 14: The long-run growth rates of creative employment and the UK workforce 1981 to 2001 from census data and 2002 to 2006 from LFS data.
• Table 15: Comparison of CCI employment estimates with those in DCMS Economic Estimates.
• Table 16: The segment shares overall of creative employment for 1981 to 2006 and long-run growth rates of employment.
• Table 17: The Creative Income Trident: total value of annual earnings generated through creative workforce salaries and wages for 2001 to 2006.
• Table 18: The mean annual income of the creative segments and the modes of employment for 2006.
• Table 19: Comparison of mean incomes of creatives with the UK mean in 2006.
• Table 20: Annualised growth rates over 2001 to 2006 of employment, earnings and mean income of the creative segments and the UK workforce.
• Table 21: UK employment as at 2001 census by Industry Division.
• Table 22: Looking within Division K: The extent of specialist and embedded employment.
• Table 23: The change in the embedded share of employment by Division between 1981 and 2001.
• Table 24: The change in the embedded share of employment by Division between 2001 and 2006.
• Table 25: The UK Financial Employment Trident for 2001.
• Table 26: Employment within the UK Financial Employment Trident from 1981 to 2001.
• Table 27: 2001 UK Financial Employment Trident by Industry Division.
• Table 28: Comparison of financial and creative embedded shares by Industry Division.
• Table 29: Comparison of the number of industry classifications in use in Australia, UK and USA.
• Table 30: Comparison of the number of occupation classifications in use in Australia, UK and USA.
• Table 31: Core creative industries selected from those used in the 2001 census.
• Table 32: Core creative industries selected from those used in the 1981 and 1991 censuses.
• Table 33: Core creative occupations selected from those used in the 2001 census.
• Table 34: Core creative occupations selected from those used in the 1991 census.
• Table 35: Core creative occupations selected from those used in the 1981 and 1991 censuses.
• Table 36: The number of people with second jobs and main jobs in creative occupations for 2006.
• Table 37: The number of people with second jobs and main jobs in creative segments for 2006.
• Table 38: The ratio of the number of people with second jobs to those with main jobs in creative occupations and industries for 2006.
• Table 39: The number of people employed in main and second creative jobs by occupation and industry segments for 2006.
• Table 40: The allocation of creative occupations across two disaggregated industries.
• Table 41: Disaggregating 7420 Architecture and Engineering Services from 2001 to 2006.
• Table 42: The number of classifications available in the 2006 Australian census for measuring the creative segments.
• Table 43: The impact of the change in classifications on the measurement of creative employment in Australia.
• Table 44: The top ten occupations within each candidate creative industry using 2001 census classifications and data.
• Table 45: The top ten industries for each candidate creative occupation using 2001 census classifications and data.
• Table 46: The top ten industries for the '2126 Design and development engineers' occupation using 2001 census classifications and data.
• Table 47: The top ten industries for the '5499 Hand Craft occupations not elsewhere classified' occupation using 2001 census classifications and data.
• Table 48: Total Australian creative employment 1996, 2001 and 2006 and the 10-year annualised growth rate.
• Table 49: Embedded employment's share of creative employment by segment between 1996, 2001 and 2006 and the 10-year annualised growth rate.
• Table 50: Australian mean annual income of people within the Creative Trident compared with the mean of their Division of employment.
• Table 51: The rate of self-employed in ‘Main Jobs' in UK creative industries 2006.
• Table 52: The correlation between employment in the 2001 LFS four-digit industry classification and the 2001 census classification.
• Table 53: Occupations included in ‘1134 Advertising and public relations managers' and '2131 IT strategy and planning professionals'.
• Table 54: The more detailed industry classifications within '72 Computer and related activities' available from LFS datasets.
• Table 55: The summary disaggregation matrix for '72 Computer and related activities'.
• Table 56: The blend of occupations employed within '742 Architectural and engineering'.
• Table 57: The summary disaggregation matrix for '742 Architectural and engineering'.
• Table 58: Creative segments' specialist, support and embedded employment for 1981.
• Table 59: Creative segments' specialist, support and embedded employment for 1991.
• Table 60: Creative segments' specialist, support and embedded employment for 2001.
• Table 61: Creative segments' specialist, support and embedded employment for 2006.
• Table 62: The 25 sub-divisions with the highest proportion of embedded employment 2001 UK census data.
• Table 63: Growth in embedded creative employment across the industry sub-sectors.
• Table 64: The changes in the creative embedded share of two-digit industries' employment between 2001 and 2006 using LFS data.
• Table 65: The classifications selected in the Cultural Industries definition and the number of people employed in 1981, 1991 and 2001.
• Table 66: The classifications selected in the Cultural Occupations definition and the number of people employed in 1981.
• Table 67: The classifications selected in the Cultural Occupations definition and the number of people employed in 1991.
• Table 68: The classifications selected in the Cultural Occupations definition and the number of people employed in 2001.
• Table 69: Cultural Trident in 1981 for each segment and mode.
• Table 70: Cultural Trident in 1991 for each segment and mode.
• Table 71: Cultural Trident in 2001 for each segment and mode.
• Table 72: Long-run annualised growth in Cultural Trident employment between 1981 and 2001.

List of figures

• Figure 1: DCMS estimates for 1998 to 2006 of sectoral employment within the creative industries and total employment.
• Figure 2: Frequency of discrepancy between census and LFS employment – industry classifications.
• Figure 3: Frequency of discrepancy between census and LFS employment – occupation classifications.
• Figure 4: The variation in the distribution of employment within the weekly income bands for each of the creative segments compared with Australian employment.
• Figure 5: The self-employment rate of creative occupation segments in 2006.
• Figure 6: The self-employment rate of disaggregated creative occupation segments in 2006.
• Figure 7: Comparison of the CCI employment estimates for each segment with those of the DCMS Economic Estimates.
• Figure 8: Comparison of the shares of segment employment estimates from CCI with those of the DCMS Economic Estimates.
• Figure 9: Comparison of the average annual incomes of specialist, support and embedded creatives within each segment with that of the national average for 2001 to 2006.

Part 1: Purpose and structure of report

Mapping studies are valuable tools for policymakers and industry professionals alike. They are a vital resource for the communication of key concepts and data. They may also potentially be used to construct an evidence base to inform government investment.

The development of a robust methodology is critical to achieving these aims. Undervaluing a sector most obviously affects its ability to secure supportive policy measures or to attract additional external investment. Overstating the sector's significance by, for example, adopting a loose definition of its activities is potentially no less damaging if it leads to a loss of credibility that turns policymakers off.

Improving the methodology used to construct estimates of the size of the UK's creative economy is a topic of ongoing interest to policymakers, driven in part by the Creative Economy Programme of the DCMS in the UK. In 2007, two major published studies have focused on this area: Frontier Economics' Statistical Analysis of the Creative Industries, ¹ which develops a more nuanced definition of the industry than in the current official statistics; and GLA Economics' measurement work for London's creative economy, ² which examines the characteristics of the capital's creative sector in the broader national context.

1.2 The Creative Trident methodology

This report complements these studies. It applies the Creative Trident methodology (defined in brief in Part 2 and in detail in Part 4) developed by the Australian Research Council Centre of Excellence for Creative Industries and Innovation (CCI) in an attempt to address some of the methodological challenges that have arisen in previous approaches to measurement of creative activities, industries and employment.

1.3 The structure of the report

Since the DCMS introduced the concept of the creative industries in 1998, various measurement frameworks and quantitative techniques have been employed in different countries to assess the extent and importance of creative activities.

The proliferation of studies has led to numerous, often conflicting, definitions of what counts as 'creative' in the economic context. To clarify our parameters, Part Two of this report sets out the definitions, assumptions and terminology we employ in our application of the Creative Trident methodology to UK data.

Part Three provides an overview of existing measurements of the creative industries in the UK and elsewhere, and analyses the strengths and weaknesses of their underlying methodologies. Part Four describes the Creative Trident methodology and compares it with existing approaches.

UK census data are in many ways limited compared with the data collected in such countries as Australia, New Zealand, Canada and the United States. For this reason, our analysis overlays data from the UK Labour Force Survey (LFS) from 2001 to 2006 on the census data available for 1981, 1991 and 2001.

Part Five reports the results of the UK Creative Trident mapping exercise. It presents the 'Employment Trident', which incorporates: data on specialist, support and embedded creative employment for 1981, 1991 and 2001; growth rates and average levels of earnings for specialist, support and embedded creative workers from 2001 to 2006; and the distribution of embedded creative occupations across non-creative industries for 2001 to 2006. Growth in creative employment and earnings is compared with economy-wide averages for the relevant periods.³

Part Six demonstrates the application of the Trident methodology to the financial services sector, and compares the extent to which financial activities and creative activities are embedded in the UK – that is, employment of creative and financial services professionals outside their respective specialist sectors.

This is particularly interesting in the light of recent studies which have indicated that the creative industries are now as large and add as much economic value as the financial services sector. This degree of 'embeddedness', we propose, could be a proxy for the extent to which the creative industries enable economic development, as the financial services sectors are often said to do.⁴

Part Seven concludes by summarising the main findings and implications arising from this study.

Part 2: Terminology and assumptions

This section describes the main terms used in this report and lists the most important assumptions that underpin the Creative Trident analysis.

2.1 Terminology

Creative industries: At a general level, the 'creative industries' is the collective term for those businesses in the economy which focus on creating and exploiting symbolic cultural products (such as the arts, films and interactive games), or on providing business-to-business symbolic or information services in areas such as architecture, advertising and marketing and design, as well as web, multimedia and software development.

In a practical sense the creative industries are defined by a selection of Standard Industrial Classification (SIC) codes that are implemented in national datasets that encompass the specialist businesses that produce creative goods or services. So, for example, the DCMS definition focuses on those activities which have their origin in individual creativity and which have the potential for wealth creation through the generation and exploitation of intellectual property.

Table 4 lists the industry descriptions and associated SIC codes for the 11 industries selected for this analysis of the 2001 household census. We will discuss later how these categories align with the 13 sectors that make up the official DCMS measure of the creative industries.

Unfortunately there is no single internationally accepted classification scheme for industry activities, although most recently there has been a degree of convergence into two main families of classification (North American and European).

However, in the context of this study, the Office for National Statistics (ONS) has used different versions and levels of detail in its implementation in the SIC scheme to code industry activity in each of the census years and in the LFS. Appendix 1 lists the various industry classifications considered to be creative industries and that are relevant to our datasets.

Creative occupations: 'Creative occupations' are a selection of occupations which produce creative goods or services, drawn from the various releases of the UK Standard Occupational Classification (SOC) codes. Individuals engaged in creative occupations may participate in any stage of the production process, but it is their involvement primarily in creative functions (rather than, for example, retail sales) that distinguishes them.

Table 4: The 11 creative industries selected from the 2001 census data.

Table 5: Creative occupations selected from the 2001 census data.

The 26 four-digit SOC creative occupations selected from the 2001 census data are detailed in Table 5.

The occupation classification scheme used to code occupational data from the earlier census years (1981 and 1991) differs from that used for the 2001 census and the Labour Force Survey (LFS). Appendix 1 contains further information.

Creative Trident: The Creative Trident comprises all the people working in the creative industries (as defined above) and the people working in specialist creative occupations in 'non-creative' firms and organisations.⁵

Trident modes: The three Trident modes are:

• Specialist mode: Those people in defined creative occupations employed within the defined creative industries.
• Support mode: Those people employed within the defined creative industries who are not working in defined creative occupations but who perform sales, management, secretarial, technical, accounting and administrative functions.
• Embedded mode: Those people employed in defined creative occupations but who are working outside the defined creative industries.

Creative Trident methodology: This is a methodology to analyse detailed data of the number of people employed, and where possible their mean income, in every occupation across every industry. The methodology aggregates the number of individuals employed in specialist, support and embedded modes and provides detailed counts according to the level of occupational and industry detail available.

Creative workforce: The creative workforce is the group of people employed in the Creative Trident (i.e. in specialist, support and embedded modes) across all selected creative occupations and creative industries. The size of the creative workforce is determined using the Creative Trident methodology.

Creative segment: A subset of the creative workforce grouped by broad categories of creative output, comprising specialist, support and embedded modes.

The original DCMS Creative Industries Mapping Document, published in 1998, defined 13 sectors, some of which could not be quantified using national data sources. The DCMS subsequently published estimates for these sectors in the broader 11-sector classification

Table 6: The correlation between CCI segments and DCMS Economic Estimates sectors.

It is important to note that, because the Creative Trident accounts both for activities within the creative industries and embedded creative occupations in non-creative industries, our definition of 'creative segment' includes both the subset of creative industries related to each segment above and the subset of creative occupations producing similar types of creative output, even where they are found in non-creative industries (i.e. as embedded employment).

For example 'publishing' is about more than the 'publishing industry' (SIC 221) alone. The creative segment 'Publishing' also refers to specialist and support employment within 'Library, archives, museums and other cultural activities' (SIC 925) as well as embedded employment (i.e. in non-creative industries) of the following creative occupations:

• 2451 Librarians
• 2452 Archivists and curators
• 3412 Authors, writers
• 3431 Journalists, newspaper and periodical editors
• 4135 Library assistants/clerks
• 5421 Originators, compositors and print preparers

2.2 Assumptions

The principal data assumptions underlying the analysis described in this report are as follows:

• That the average income of those who are self-employed within each occupation-industry combination is the same as those working as employees. If the mean income of the self-employed is understated or overstated by 50 per cent, say, from the true mean then this would result in an under- or overstatement of annual creative earnings in our estimates by up to 14 per cent.
• That LFS scaling is accurate: LFS is a 1-in-400 sample, and results reported here have been scaled up by the Office for National Statistics (ONS) to produce population estimates. We provide no independent check on the accuracy of the ONS's scaling-up process. (See Appendix 2 for a comparison of LFS and census employment).
• That census scaling for 1981 and 1991 is appropriate: the ONS collected employment-related data for the censuses held in these years on only a 10 per cent sample of the overall census population. We have to assume that the ONS's methodology ensures that this 10 per cent is representative, so that we can scale up results for these years by a factor of 10 to produce a population estimate.
• That UK creative incomes are meaningfully represented by mean weekly incomes when determined for each occupation employed within each industry at the finest level of classification available.
• That by excluding creative employment as a 'second job', we have not introduced an appreciable bias into our estimates within the LFS dataset. Comparisons of creative employment to the total workforce can only be made on the basis of employees' 'main job'. See Appendix 3 for an analysis of the impact of excluding second jobs.

Part 3: Existing analyses of creative activities

The first Creative Industries Mapping Document by the DCMS⁶ became the template for numerous other studies commissioned by governments at the national, regional and city level. The DCMS study built on many earlier attempts to study the size and impact of the cultural industries, and established the approach of measuring employment and business activities within selected industrial classifications, using data from household surveys of the industry in which individuals work, or from surveys of businesses within specific industries.

Other mapping studies, such as by the Hong Kong Creative Industries Baseline Study, ⁷ refined the early DCMS template to develop a more comprehensive approach including examination of specific creative occupations.

Subsequent analyses have further attempted to measure employment in the creative or cultural workforce, as exemplified by the French Culture Ministry's Department for Planning and Statistics 2005 report, ⁸ work conducted for Australia by the CCI authors, ⁹ and research conducted for the UK by Frontier Economics as part of the 2007 Creative Economy Programme. ¹⁰

The development of a robust measurement methodology is critical to achieving the primary function of mapping studies. This section of the report summarises the various approaches used to map the creative industries and creative economy activities in the UK and in other jurisdictions, and highlights the strengths and weaknesses of each.

3.1 Industry-based studies

3.1.1 The 1998 and 2001 DCMS Mapping Documents

The 1998 UK Creative Industries Mapping Document, produced by the DCMS, identified 13 creative industries for analysis: Advertising; Architecture; Art & Antiques Market; Crafts; Design; Designer Fashion; Film & Video; Interactive Leisure Software; Music; Performing Arts; Publishing; Software & Computer Services; and Television & Radio. It proposed that the creative industries "are those industries which have their origin in individual creativity, skill and talent and which have a potential for wealth and job creation through the generation and exploitation of intellectual property".

Measures of economic activity analysed in the Mapping Document included: levels of employment; characteristics of the firms in the identified creative industries (including their number, their size, turnover and profit margins); and outputs, including the value of exports and estimates of Gross Value Added.

A template for analysing the creative industries was formalised in the subsequent 2001 UK Creative Industries Mapping Document, and this structure was applied in New Zealand in 2002 (Creative Industries in New Zealand: Economic contribution¹¹), in Australia the same year (Creative Industry Cluster Study Series¹²) and Singapore (Economic Contributions of Singapore's Creative Industries¹³) in 2003. It was applied at a regional level, in Queensland, Australia (Creativity is Big Business¹⁴) in 2004.

3.1.2 Limitations in DCMS Mapping Documents

The DCMS template-based studies were significant in establishing the first comprehensive benchmarks for the creative industries. However, there were three key methodological limitations:

1. Conceptual limitations
2. Classification and granularity limitations
3. Data source limitations

These limitations made it difficult to compare the performance of creative industries over time, or to measure comparative performance between industries and regions.

Conceptual limitations of the DCMS template arose primarily from ongoing technological progress and changing market and industry boundaries. In particular, the progressive convergence of the information technology, communications, cultural and content industries created substantial difficulties for those trying to measure and understand economic and industrial activity.

Old occupations and sectors had emerged, especially in the digital industries, and these were not reflected in the standardised classification schemes used by statistics organisations. This lack of alignment was exacerbated by the usual 10-15 year gap between the updates to industry and occupation classifications.

A segment-based approach requires that segment-specific estimates of economic activity can be summed to provide an estimated total for the creative industries as a whole. A risk of double-counting is incurred where the results of independent, ad-hoc surveys are combined without making adjustment for any overlap in respondents or domains of interest. Unfortunately the degree of any such overlap is difficult, and sometimes impossible, to determine, as the identified survey responses for each segment are not available for reconciliation.

3.1.3 The DCMS Evidence Toolkit

The 2002 UK DCMS Regional Cultural Data Framework¹⁵ was a substantial effort to address the need for consistency by identifying industry activities and occupations along a value chain in each creative segment. Five years after its introduction, it remains the most rigorous tool for creative economy performance tracking yet devised.

The generic value chain established by the Framework identifies the following stages:

1. Creation
2. Making
3. Dissemination
4. Exhibition/Reception
5. Archiving/Preservation
6. Education/Understanding

However, this generic value chain does not differentiate the distinctively 'creative' activities involved in the production and distribution of creative goods and services. Specifically:

• Making: under the Framework, this stage includes many manufacturing activities such as 'manufacture of relevant capital goods', which includes the manufacture of musical instruments, computers, television sets and radio transmitters. While these are important economic activities, they do not in themselves constitute content creation, nor do they require specifically creative skills.
• Dissemination and Exhibition/Reception: under the Framework, activities conducted by entities such as cinemas and newsagents are included. However, these are dominated by the distribution of creative outputs generated in other countries. For example, in the UK film market, UK-originated content accounts (including US co-productions) for only around 20 per cent of all cinema box office receipts and 17 per cent of all DVD and VHS rental transactions. ¹⁶ Similarly, newsagents typically gain little revenue from the sale of newspapers, and their magazine and book trade could well be dominated by non-UK titles.

While accounting for these types of activities may be appropriate when developing 'access to markets' strategies or conducting input/output and economic multiplier studies, we suggest that they are not directly relevant when calculating the level of creative employment.

3.1.4 Capturing creativity

Defining what does count as 'creative' however is a challenging task. The segment definitions developed in the 1998 DCMS template are based on the common types of creative output within industries; studies using the template therefore analysed data about the specialist firms operating within each specific segment.

However, as Pratt (2004)¹⁷ notes, standard industrial classification schemes are not good at describing creative activities, which are generated in three ways: first, by individuals (as sole traders or producers); second, by groups working within specialist (often consulting) creative organisations; and third, by individuals in creative occupations working in non-creative organisations. ¹⁸

The substantial and frequent movement between these modes of activity exacerbates the limitations of standard industry classifications. For example, individuals may move from being an independent film producer to working for government film agencies, or a designer might sign a three-year contract to work for a bank or advertising agency.

A result of this movement and diversity is that the economic significance of creative activities may be underestimated if one relies on an industry-centric approach. Measuring the employment significance of, for example, the software segment, would involve counting the people employed within firms in the specialist computing software industries under the most appropriate classification (in this case, '72.2 Software consultancy & supply'). Australian evidence¹⁹ suggests that this approach has led to substantial under-counting in that country by approximately 50 per cent, reflecting both poor statistical industry definition coverage and the fact that so many software designers are 'embedded' in other industries.

Another practical classification issue is that many creative activities are 'lost' within broad industry categories where they cannot be included without significant risk of overreach. For example, in the UK, specialist design consulting activities are buried within broad industry classifications such as '748 Miscellaneous business activities not elsewhere classified' or '222 Printing and service activities related to printing' and even in several classifications related to clothing manufacturing.

This lack of alignment arises when industry data are not classified with sufficient detail to enable specific measurement of creative activities. Where the relevant creative activities are subsumed into other industries, estimates are often based on a proportion of the larger industry. The DCMS's Economic Estimates (DEE) series, for example, attributes 5 per cent of a range of clothing manufacturing activities to "Designer fashion". Such factors are subject to adjustment in the light of better knowledge or changes in the workforce, and result in the need for revisions of earlier estimates.

In other cases, such as the UK 'Design' sector (which comprises Graphic Design and Industrial or Product Design), there are currently no UK SIC codes used to categorise national survey data and they are consequently silent on the subject.²⁰

The coding of 'industry of employment' responses in the 2001 UK census resulted in only 11 classifications being useful at the three-digit SIC level and one ('Computer and related activities') at the two-digit SIC level, compared with 34 in Australia. ²¹ The accuracy of estimates of creative industries employment in the UK would be substantially improved if the data were compiled at the four-digit resolution available in other countries. These data limitations have meant that the DCMS's industry-led estimates of employment have not made use of UK census data, with the result that the DEE are calculated solely on sample-based surveys – primarily the Labour Force Survey - and where necessary industry-specific business surveys.

3.2 Capturing industry and occupation

From 2003, a number of new creative industry mapping documents²² looked beyond industry analysis, and began to explore occupational analysis, counting the number of people employed in specialist occupations and the patterns of employment across creative sectors.

The DCMS²³ estimates that creative jobs outside the creative industries add approximately 800,000 people to the estimated 1.1 million working in the creative industries. Figure 1 plots estimates of the UK's total creative employment and, within that, employment in each of the DCMS-defined creative sectors.

Figure 1: DCMS estimates for 1998 to 2006 of sectoral employment within the creative industries and total creative employment.

The occupational data were extracted from the annual Labour Force Survey a panel-based survey of 350,000 individuals in the UK which is then scaled up.

Source: DCMS Economic Estimates

3.2.1 Methodological constraints of the approach

The methodology of combining occupation and industry data was further developed in the Hong Kong Baseline Study and the Ontario Design Study, especially in the refinement of methods for establishing patterns of embedded employment.

While the inclusion of creative occupations has been an important step forward for creative industries mapping studies, methodological and data limitations have continued to limit the accuracy of these approaches. The diversity of creative activities makes it important for analysis to be conducted at the finest possible resolution of classification.²⁴

The DCMS Economic Estimates are constrained by either having to use lower-resolution classifications which lead to greater margin of error, or to use tables with data selected at higher resolution classifications but only supplied by the statistical authorities within aggregated industry totals. The preferred choice, obtaining detailed occupations along one axis and detailed industries along the other, is not feasible as it results in the suppression of a significant proportion of the LFS source data in individual cells of the table. ²⁵

The impact of this for the DCMS is that it cannot perform analysis at a fine level in more than one dimension (viz. occupation within industry) at any reasonable level of detail, as doing so would generate substantial statistical errors. DCMS's approach has been to obtain summary cross-tabulations of the level of employment in each of the selected creative occupations that occurs outside the defined creative industries.

This limitation has several consequences:

1. Estimates of the embedded creative workforce are calculated as a residual of total occupations and industry employment, and so depend on what creative industries are selected.
2. It is not possible to analyse the distribution of creative employment in businesses outside the creative industries. So, for example, the published summary tables do not allow researchers to determine either the significance of the top ten occupations within a selected industry, or the top ten industries in which a selected occupation is employed.²⁶

3.2.2 Classification selection and embedded employment

Care also needs to be taken in selecting which industry and occupation classifications to analyse, as the chosen candidates should vary, depending on the purpose and context of the classification. The industry and occupation classifications used to calculate employment for the DEE appear to have been selected without considering how these categories interact in multi-dimensional, whole-workforce datasets.

Three factors need to be considered when analysing combined industries and occupations datasets to determine the level of embedded employment:

Correlations between occupations and industries

The DCMS Economic Estimates 2006 report includes four occupations in the Publishing segment which certainly have a place within the publishing industries, but could not be considered as creative occupations when embedded within the broader economy. These include 'Originators', 'Compositors and Print Preparers', 'Printers, Bookbinders and Print Finishers' and 'Screen Printers'. Of the 35 occupation codes used by the DCMS to calculate embedded creative employment, ten are arguably largely or wholly irrelevant, and substantially inflate estimates of the 'embedded' creative workforce.

Furthermore, of the 25 industry classifications used by the DCMS in the calculation of industry employment, export and business numbers, six are likely to have a very low degree of correlation with core creative activities. These include 'Clothing Manufacture', 'Other Business Activities Not Elsewhere Classified', 'Other Retail Sale in Specialised Stores', and 'Retail Sale of Second-hand Goods in Stores'. Although proportional adjustments are made to some of these industry data to reflect the fact that the industries are not wholly creative in nature, such scaling adjustments create other problems (see 'Maintaining the integrity' below).

Scaling obscures the details

The application of adjustments to broader categories of activity hampers the ability to reveal outliers in the employment data or changes in circumstances. In addition, the current scaling factors have remained fixed since the analysis was first conducted in the 2001 DCMS Mapping Study. Given the rate of change and convergence in the creative industries in recent years, it is unlikely that these proportions can have remained relevant without some further adjustment (as recognised by the DCMS).²⁷

Maintaining the integrity of a whole-of-workforce dataset

The calculation of embedded employment cannot be simply grafted onto a methodology intended to determine the extent of employment across every activity relevant to the creative industries.

The 2001 census shows the total employed population in the UK as 26,575,775 people, analysed on the basis of either occupations or industries. The LFS 2006 dataset shows 28,165,612 people employed (on the basis of main job) on either axis. Applying a scaling factor to one axis, such as the industry classification axis, is not straightforward as doing so invalidates the integrity of the two axes' totals.

If the intention of an employment analysis is to show the size and characteristics of a segment in the context of the whole of the workforce, then irrespective of whether the research is using detailed (as in the census) or summarised (as in LFS sub-totalled) two-dimensional tables, a proportion of an industry cannot be used without creating an imbalance in the axes' totals. In this respect, working with an 'occupation within industry' employment table is rather like double-entry accounting – it is more onerous to establish and maintain, but the reward is much greater flexibility in reporting and the improved security of built-in integrity checks.

Part 4: The Creative Trident methodology

The CCI Creative Trident methodology was developed to address the methodological challenges described above and so provide a more robust foundation for meeting the requirements of mapping studies. ²⁸

CCI has developed a methodology and suite of tools which enable analysis of a range of national datasets (such as census-based employment and business activity) from different periods and sources, despite variations in industry/occupation/qualification classifications, often with different underlying hierarchical strategies and varying levels of detail aggregation.

Our previous research has shown that the significance of creative activities – their level of employment and earned income – can be determined using a methodology centred on a census-based matrix combining employment within the creative industries and creative employment within other sectors of the economy. Our methodology attempts to combine these data in a way that neither inappropriately includes nor excludes people because of shortcomings in classification schemes.

The Trident definition of creative activity therefore recognises three distinct employment situations, or 'modes':

• Workers within a creative profession (determined by occupations) within a creative sector (determined by industries) 'specialist' mode.
• Workers in a non-creative profession within a creative sector 'support' mode.
• Workers in a creative occupation outside the creative industries 'embedded' mode.

These three employment modes together comprise the Creative Trident. The Trident can be analysed from three perspectives: employment (disinguishing between specialist, support and embedded workers); income (deriving the total annual incomes generated by specialist, support and embedded workers activities); and whole-economy (disaggregating the extent of embedded employment across each non-creative sector of the economy).

4.1 Defining the creative core

The Creative Trident methodology can be applied using any well-articulated definition of activities (for example, creative, cultural or financial services). However, it works best conceptually when there is a concentration on what we call the 'pre-creation and creation' stages of the value chain, which we refer to collectively as the 'creative core'. Concentrating on these stages means that we capture the essential starting points for creative activity, whether in the creative industries themselves (the specialist mode) or in the wider economy (the embedded mode).

While this is a different approach from that of the DCMS Economic Estimates, it employs a selection strategy very similar to Layer One of the five-layer generic supply chain concept proposed by Frontier Economics in its recently published work for DCMS.²⁹ Frontier suggests that the level of creativity declines from Layer One to Layer Five, and that Layer One is the most appropriate to use for most benchmarking purposes.

Our definition of the creative core³⁰ selects the activities, in either occupation or industry classifications, which occur at the pre-creation stage (including preservation, access, collecting and licensing activities), ³¹ and the creation stage of the value chain. In the creation stage, we follow David Throsby's notion of 'creative workers', defined as: ³²

• those engaged in producing primary creative output - for example, writers, musicians, visual artists, film, television and video makers, sculptors and craftspeople;
• those engaged in interpretive activity – for example, performers interpreting works of drama, dance, music etc. in a wide variety of media from live performance to digital transmission via the Internet; and
• those supplying creative services in support of artistic and cultural production – for example, book editors, lighting designers, music producers, etc.

While Throsby's definition is essentially occupation-focused, deriving as it does from cultural employment, it may also be applied to industry-defined activities and services.

We have excluded a number of DEE industry activity selections: some in which the DCMS counts all employment (such as printing, film distribution, cinemas, and the provision of specialist education services), and some in which they apply proportional calculations (such as the production of those capital goods e.g. TV sets and musical instruments which may be needed for creation, performance or usage).

By excluding these, we create a basis for more reliable comparisons of employment (including self-employment) between the six segments at both the pre-creation and creation stages. As with Frontier's Layer One focus, our strategy has been to focus on the creative and production-related activities which demonstrate the highest degree of creative value-added. These are also the stages where there is greater availability of classifications to measure their activities.

Applying the Creative Trident methodology requires the selection of whole occupation and industry classifications since proportions cannot be used in 'occupation within industries' datasets. Equally, apart from a few very broad, poorly defined classifications, the selection process is based on the actual patterns of employment observed in the data:

• Creative occupations typically have at least 25 per cent of their employment in creative industries.
• Creative industries typically have at least 25 per cent of employment in creative occupations.

Difficulties in the selection of classifications – or more often in their grouping into segments – become apparent when the first analysis is performed, as one segment could have little or no employment in one mode – very often in support mode. So since only 11 industry classifications are being used, it is difficult to group them into more than six segments.

If we bear in mind that we are selecting in two dimensions, not one, then the industries which meet these criteria at the most detailed level are as shown in Table 7, and the qualifying occupations are as shown in Table 8.

Table 7: Classifications used to determine the creative core industries using census and LFS data and the percentage of employment that is within creative occupations.

Source: Analysis by CCI of custom tables of the Office for National Statistics 2001 Census of Households and 2006 LFS

Under our criteria, we have included three industries excluded by the DEE:

1. 9251 Library and archive activities
2. 9252 Museum activities and preservation of historical sites and buildings
3. 3622 Manufacture of jewellery and related articles not elsewhere classified

Conversely, six of the industries included in the DEE are excluded from our analysis:

1. 2231 Reproduction of sound recording
2. 2232 Reproduction of video recording
3. 2233 Reproduction of computer media
4. 9212 Motion picture, video distribution
5. 9213 Motion picture projection
6. 9272 Other recreational activities

These industries have been excluded because they relate primarily to the distribution and exhibition of creative outputs, thus falling outside our pre-creation and creation criteria for inclusion in the creative core.

Additionally, automatically excluded from our selection are any of the classifications from which the DCMS only uses proportions of employment: the nine clothing manufacture and two art and antiques retail industries. ³³

For occupation of employment, our selection comprises the 26 SOC codes listed in Table 8.

Table 8: Creative core occupations used in census and LFS datasets and the percentage of employment occurring within creative industries in 2001.³⁴

Source: CCI analysis of Office for National Statistics 2001 Census of Households custom table and 2006 LFS

On the surface it would appear that '5491 Glass and ceramics makers, decorators and finishers' and '5492 Furniture makers, other craft woodworkers' should not be included in our estimates; but in both cases a significant proportion of the employment is in industries that should be considered at the very least marginal creative industries and so have been included: 41 per cent of glassmakers are employed in '262 Manufacture of non-refractory ceramic goods other than for construction purposes; manufacture of refractory ceramic products', while 54 per cent of furniture makers are in '361 Manufacture of furniture'.

Both occupation classifications are included because they encompass the design and realisation of custom and handcrafted glass and furniture items.

Under our criteria, five occupations are included in the creative core which are excluded by the DCMS DEE:

1. 2132 Software professionals
2. 2451 Librarians
3. 2452 Archivists and curators
4. 3122 Draughtspersons
5. 4135 Library assistants/clerks

Conversely, we exclude 14 occupations that DCMS include in employment calculations:

1. 1136 Information & communication technology managers
2. 2126 Design and development engineers
3. 3433 Public relations officers
4. 5411 Weavers and knitters
5. 5422 Printers
6. 5423 Bookbinders and print finishers
7. 5424 Screen printers
8. 5493 Pattern makers (moulds)
9. 5494 Musical instrument makers and tuners
10. 5499 Hand craft occupations not elsewhere classified
11. 8112 Glass and ceramics process operators
12. 9121 Labourers, build & woodworking trades
13. 5244 TV, video and audio engineers
14. 5496 Floral arrangers, florists

Again, our definition of the creative core excludes these occupations as they are more closely related to the later stages of the value chain manufacture and distribution and do not intrinsically require creative skills to the same extent.

Our core definition of 'creative industries', we believe, establishes a justifiable demarcation between specialist and embedded employment, while our core definition of 'creative occupations' makes the measurement of embedded employment more robust.

The effects of this more restrictive selection on total employment data are mitigated by the fact that the methodology, relying as it does on two-dimensional occupation within industry employment datasets, still counts the employment of those in creative occupations, regardless of whether or not they work in the creative industries as defined.

4.2 Advantages of the Creative Trident approach

The Creative Trident represents an advance on previous creative industries mapping approaches in five ways:

1. Avoidance of tendency to overreach.
2. Disaggregation of creative employment to measure embedded employment within non-creative industries.
3. Decomposition of specialist and support employment within creative industries.
4. Use of population-based data sources rather than surveys, whenever possible.
5. Use of annual earnings generated from survey data of the employed (including self-employed).

4.2.1 Avoidance of overreach and robustness

From a conceptual perspective, the definition used within the Trident methodology recognises that the critical value-added of creative activity occurs at the beginning of the value chain, in the creation and publication of creative works of all kinds. CCl's methodology only considers employment as relevant for inclusion where it occurs in the core creative value chain, and not in related chains such as distribution or retailing. ³⁵

4.2.2 Disaggregation of embedded employment

The Trident methodology allows analysis of the extent to which embedded employment is distributed across the economy, such as the incidence of software professionals working in the finance sector and advertising professionals working in the manufacturing sector. Understanding the distribution of creative employment on a sector-by-sector basis provides an improved evidence base for policymaking decisions.

4.2.3 Breakdown of specialist and support employment

The Trident methodology explicitly separates the employment within specialist businesses into either core or support occupations. Being able easily and consistently to determine the ratios of support and management staff to creative workers and fee-earners for the creative sectors could provide fertile ground for research into the commercial viability and productivity of creative businesses. At the very least, the disaggregation helps us to understand the creative workforce and is more robust than relying on a single dimension, whether industry or occupation.

4.2.4 Use of population data

Population-based datasets are preferable because they allow us to produce true calculations, rather than sample-based estimates, of employment and earnings. While these are still subject to the normal errors involved with census data collection, such as classification errors in the coding of census returns, any such errors that do occur should be random and should therefore not affect one sector more than another.

In the Australian context, the ability of the Creative Trident methodology to take advantage of the Australian census has allowed highly robust calculations which are more accurate than previous attempts to measure the extent of creative economy activity within Australia. ³⁶ Ongoing work by CCI indicates that results of similar quality should be obtained from application of the Trident methodology in New Zealand.

4.2.5 Use of earnings data

The salaries, fees and profits paid to staff and business owners broadly reflect the value of an activity to the economy (but not necessarily to society as a whole as this ignores, for example, spillover benefits). The ability to determine with reasonable accuracy the total personal earnings arising from creative employment therefore provides an indicator (if not a proxy) for its value-added.

Earnings data combine both the level of employment and the average level of remunerations received by individuals (specifically the mean income). The methodology allows the calculation of earnings from mean income of each combination of occupation within industry: graphic artists may earn substantially more working within a bank's corporate communications department than they would working for an education department, for example. This difference cannot be determined with single-dimension datasets (in this example, simply by taking the mean income of graphic artists or the mean income of banking staff or education department staff).

The Trident methodology not only calculates total earnings more accurately, but can also highlight some of the disparities between different roles within a segment. One such example – it turns out – is that managers, administrators and others in the support role in the Music and Performing Arts segment appear to earn significantly more than specialist musicians. (See Appendix 8 for examples of the types of analyses produced from Australian census personal earnings data.)

4.2.6 Applying the Creative Trident to UK data

A key strength of the Creative Trident methodology is that once the foundation analytical table has been produced from a processed dataset, a number of different analyses can be performed, each of which can reveal a different aspect of the creative workforce. Because of the reliability and coverage factors, the methodology would ideally be applied to a dataset from the full population. This would allow multiple dimensions – most importantly, the occupation for each industry of employment at full resolution, preferably within a third dimension of income bands.

The more industry and occupation classifications that are used at the detailed level, the greater the ability of researchers to distinguish activities relevant for the analysis from those that are not. In Australia, the 2001 census dataset held 1,300 occupation classifications with 500 industry classifications and with income coded within 16 bands. This resulted in over 400,000 cells with employment counts.

While census data are preferable because they avoid the small-sample and selection bias issues that can arise through sampling, some countries, such as the USA, conduct labour force surveys with good sample rates: the 1.2 million businesses included in the US sample ensure that there is almost no suppression applied to low count cells. ³⁷

However, the limitations in the available UK data substantially constrain the robustness and accuracy of quantitative analyses of creative employment, including the Creative Trident.

The overarching limitations of the data include:

• no single dataset reliably provides the basic information required; and
• limited coverage of the self-employed.

Specific limitations of the census data include:

• ten-year frequency of the UK census;
• low resolution of classifications, especially by industry; and
• the UK census's exclusion of individual incomes.

The major limitations of the LFS from our viewpoint are that:

• the survey basis constrains multi-dimensional analysis to sub-totals;
• there is improved but still low resolution of classifications, especially by industry;
• no income data are available for the self-employed; and
• there are inconsistencies within the census employment data at detailed levels.

In addition, the full LFS datasets are not straightforward to acquire and they required considerable processing prior to analysis – in marked contrast to the availability and research readiness of other countries' employment datasets (many of which are free to download).

The following sub-sections examine each of these limitations in turn.

4.3 Overarching UK data limitations

4.3.1 No single dataset provides the basic information required

The minimal requirements for applying the Creative Trident cannot be met by a single data source in the UK: there is no full-resolution census table of employment and income for each occupation within each industry for both employees and the self-employed.

In addition, unlike the censuses in the US, Canada, New Zealand and Australia, the UK census has not historically collected data on individual incomes. Given this limitation, the Labour Force Survey data has to be used to estimate the Creative Income Trident.

4.3.2 Limited coverage of the self-employed

Additional data sources are also required to get a full picture of self-employment within the UK creative sector. Although the census covers all employment configurations, it is important to consider the coverage of self-employed individuals in other data collections, too, as the self-employment rate for the creative industries in the UK is as high as 27 per cent, and for creative occupations, 28 per cent. ³⁸ These rates are double those found in the UK economy as a whole (13 per cent). (See Appendix 9, Self-employment rates within the creative occupations and industries.)

Apart from the census, the Labour Force Survey is the only relevant UK data source which provides full coverage of the self-employed.

The Annual Survey of Household Earnings (ASHE) does not survey the self-employed, since inclusion in the sample requires that the individual participate in the 'Pay as You Earn' tax scheme. The Annual Business Inquiry (ABI) does not survey non-employing enterprises, thus excluding a potentially significant number of sole-practitioner specialist creatives.

This limited coverage of the self-employed explains why we use the LFS as the sole secondary data source for adjustment and extension of census data.

4.4 Census data issues

4.4.1 Ten-year frequency of UK census

The UK census is conducted every ten years, compared with the five-yearly censuses in Canada, Australia and New Zealand. The US also has a ten-year census cycle but conducts large scale labour force surveys every three years which reduces the impact of the longer interval. The ten-year UK frequency obviously reduces the relevance of census data, especially during periods of rapid economic change and for industries subject to significant technological disruption and convergence.

The effect of these factors is exacerbated by the fact that the research repeated here is being undertaken past the mid-point of the ten-year interval; the data gathered during the most recent census are already almost seven years out of date. However, the Creative Trident methodology could help NESTA and other interested parties to lobby for improvements to the next UK census in 2011 (see Part 7, Main findings and implications).

4.4.2 Low resolution of industry and occupation classifications

The UK census does code individual occupations at a consistent and reasonably fine (i.e. four-digit) resolution, but industries of employment are coded at a much lower level, with the bulk of results classified at the three-digit level, and even a few at two digits. And few of those activities available at four-digit resolution are related to creative activities. In Australia, by contrast, census data relating to industry of employment are consistently coded at the four-digit level (and in the case of the Australian Business Register, at five digits).

Of the 172 detailed industry classification codes used in the UK 2001 census, only nine are directly relevant to the creative industries, with a further two being included after they were disaggregated (Table 4). This is low when compared with the 28 industry classifications that are available for use on Australian datasets.

Of equal concern is the variation in the resolution at which census data are classified; some industries, such as '72 Computer and Related Activities' are only available at two digits of detail, even though they account for the employment of over 540,000 people, or 2 per cent of the UK workforce; in contrast, several general manufacturing classifications include fewer than 500 people employed. A similar issue exists with '742 Architecture and Engineering', which accounts for the employment of 390,000 people spread between many disparate fields. The situation is a little better when selecting on the basis of occupation of employment: of the 354 occupation classifications used in the UK census, only 26 are directly relevant to creative occupations (Table 5), compared with 85 in Australia and New Zealand.

4.5 Labour Force Survey dataset limitations

4.5.1 Survey basis results in the suppression of much of the detailed data

Partly to reduce the usage of possibly erroneous data, the ONS does not release LFS data in cells that hold (scaled) counts of fewer than 2,000 people. Unlike the US Bureau of Labor Statistics, with its sample size of some 1.2 million, a complete dataset of the UK LFS is unavailable to a researcher, who must work with sub-totalled cross tabulations for the creative industries and creative occupations.

At the detailed level, in a matrix with 17 rows and 28 columns, only 82 of the 476 cells contain a value, while a further 80 contain a value that is suppressed. Even entries with fewer than 2,000 people may have a significant impact at the aggregate level. The sample constraints also decrease the reliability of the mean income data for cells with lower employment counts.

4.5.2 Improved but still low resolution of classifications, especially industry

The LFS dataset still has only 18 industry classifications which fall within the CCI'S definition, and one of these (Architecture and Engineering) is very broad-based, with almost equal numbers of people in non-relevant technical consulting and engineering occupations as there are in the relevant architectural occupations (Appendix 4). It is expected that the introduction of the SIC 2007 for 2008 data collection will improve this situation with up to 35 classifications being relevant if it is implemented to the fullest level of detail.

4.5.3 No income data available for the self-employed

As noted above, this is a major shortcoming that creates the potential for error to the extent that income patterns of the self-employed differ systematically from those of creative businesses' employees.

4.5.4 Lack of comparability of LFS data with census employment at the detailed levels

Comparisons at the most detailed level between the 2001 census data and the corresponding Labour Force Survey data (see Appendix 2) reveal a substantial difference in the numbers of people employed in most of the relevant industry and occupation classifications which can be as high as 20 per cent to 30 per cent, despite the fact that the total employment figure shows only a 1 per cent discrepancy between LFS and census data.

Reliably measuring creative employment requires the addition of a limited number of detailed classifications. On the assumption that there are no differences in the level of accuracy of the coding of LFS and census responses to the appropriate occupation and industry classifications, then a census should be a more accurate representation of the employment situation than a labour force survey because it has no sampling or scaling biases.

Significant variations in the employment levels of some classifications between the census and LFS could indicate that the current LFS is less than ideal for establishing the level of employment of creative segments.

It would be worth determining the extent of variation between LFS and census data and whether there are ways to reduce the discrepancies.

4.6 Methodology to address UK data issues

4.6.1 2001 census dataset

The 2001 census dataset utilises only 173 classifications at the detailed level, of which only 11 were usable within the Creative Trident industries definition (reproduced from Table 4 below).

Table 9: Creative industry – relevant classifications from the 2001 census.

The list above was determined by analysing each industry classification in (up to) three ways:

1. By referring to the classification coding index that is used when processing a census or LFS return (see Appendix 11 for an explanation of coding lists).
2. By referring to a table generated from the census data that shows the proportion of occupations employed within the industry classification (see Appendix 6) looking for industries where the creative occupations typically account for more than 25 per cent of the industry's employment.
3. By correlating the proportions of the more detailed (four-digit) industries available in the LFS dataset to the three-digit aggregation level or, in the case of '72 Computer and related activities', to the two-digit level used within the census (see Appendix 2).

Two segments emerged from this analysis which had significant 'candidate' industry classifications, but they were deemed to encompass too many out-of-scope activities, and would require disaggregation in order to avoid distorting the results.

Consequently the number of people employed within the detailed data records for '72 Computer and related activities' was split between '7220 Computer Software consultancy' (relevant) and '7260 Other computer activities' (not relevant). In the same way, the number of people employed within '742 Architectural and engineering Services' was split between '74201 Architecture Services' (relevant) and '74209 Other Technical Services' (not relevant).

The disaggregation of occupations within industry data records is not the same as scaling an analysis of employment by industry. The process is described in Appendix 12, Census 2001: Disaggregating the data records of two industry classifications.

4.6.2 1981 and 1991 census datasets

As noted above (see section 2.3, 'Caveats'), the analysis of historical census datasets, even those only 10 or 20 years old, presents challenges when trying to establish the emergence and growth of new sectors. The standard industry and occupation classifications used to measure employment are changed infrequently, and can lag the appearance of significant new activity by up to 15 years.

Such is the case with the 1981 and 1991 censuses, in which despite implementing 310 industry classifications and thereby having 50 per cent more industry classification than the 2001 census, only 11 were considered directly relevant for our purposes. The limitations of the available industry classifications in the 1981 and 1991 census include the absence of any relevant architecture and design classifications, and the excessively broad definition of computer services, which encompass many hardware-related activities beyond the scope of our study.

Occupation classifications in the 1981 and 1991 censuses fared slightly better, with 20 out of the 373 utilised being relevant to the creative economy.

Table 10: Creative employment-relevant industry and occupation classifications from the 1981 and 1991 censuses.

4.7 Conclusion

The various limitations of the UK data discussed in this section means that the Creative Trident mapping is constrained in terms of the accuracy and robustness of results. This implies a considerably greater margin of error than the Creative Tridents calculated for Australia and New Zealand, for example. More robust estimates will require the statistical authorities to address the underlying issues of classification coarseness and coverage in the primary data sources.

Part 5: UK Creative Trident results

This part presents: the UK Creative Employment Trident, incorporating specialist, support and embedded creative employment for 1981, 1991 and 2001; growth rates in creative employment; the Creative Income Trident (the levels of creative earnings for specialist, support and embedded creative workers from 2001 to 2006); and the distribution of embedded creative occupations across non-creative industries for 2001 to 2006. Growth in creative employment and earnings is compared with economy-wide averages for the relevant periods.

5.1 UK Creative Employment Tridents: 1981, 1991 and 2001

Table 11 shows that in the 1981 household census there were approximately 900,000 people in creative employment, representing 3.9 per cent of the workforce.

Table 11: The UK Creative Employment Trident for 1981 from the census.

Source: Analysis by CCI of custom data table from the Office for National Statistics

Table 12: The UK Creative Employment Trident for 1991 from the census.

Source: Analysis by CCI of custom data table from the Office for National Statistics

Of these, almost half were employed in businesses working in the creative industries. Those working in specialist creative occupations represented only 35 per cent of the employment in creative industries.

The remaining people in creative employment – those working outside the creative industries – represent 74 per cent of those in creative occupations. However, it is quite likely that a significant degree of this 'embedded' employment is due to the limited ability of the industry classifications used over the period to capture many of the newer specialist creative business activities.

Table 12 shows that the number of people in creative employment had risen to over 1.1 million by 1991. Growth in the number of specialist creative occupations in the creative industries appears to have been particularly marked and this growth cannot be discounted as being the result of classification shifts as there was insignificant change in those over the period.³⁹

Table 13 illustrates that the total level of creative employment increased substantially between 1991 and 2001, rising to almost 1.9 million people, or 7.1 per cent of the UK's workforce. There were 1.2 million people employed in the creative industries, representing 66 per cent of total creative employment. Again, growth appears to have been particularly rapid in the number of specialists but there was also a significant increase in the number of support workers.

Table 13: The UK Creative Employment Trident for 2001 from the census.

Source: Analysis by CCI of custom data table from the Office for National Statistics

5.1.1 Growth rates in UK creative employment

Over the 20-year period between the 1981 census and the 2001 census, creative employment in the UK experienced a long-run annual growth rate of 3.8 per cent, substantially higher than the overall growth in UK employment of 0.8 per cent (Table 14).

Between these two censuses, specialist employment – that is, those in creative occupations working within the creative industries – experienced the highest growth rate of all categories, at 6.5 per cent.

Over a longer 25-year period the average annual growth rate of creative employment remained significantly higher (at 3.2 per cent) than that of the total workforce (at 0.8 per cent).

Consistent with the findings of NESTA (2007)⁴⁰ the proportion of embedded employment has actually fallen over the period from 51 per cent of total creative employment in 1981 to 34 per cent in 2001 on census data and has remained at 35 per cent since using the LFS dataset – despite the absolute growth in embedded employees.

Table 14: The long-run growth rates of creative employment and the UK workforce 1981 to 2001 from census data and 2002 to 2006 from LFS data.

Source: Analysis by CCI of custom census and LFS data tables from the Office for National Statistics

Table 15: Comparison of CCI employment estimates with those in DCMS Economic Estimates.

Source: Analysis by CCI of DCMS reports and custom census and LFS data tables from the Office for National Statistics

5.1.2 Comparison with DCMS estimates

Our calculations of the overall numbers of people in creative employment sit well with the total employment numbers modelled for the DCMS Economic Estimates (DEE), varying by as little as 2.4 per cent (2003) to 6.2 per cent (2004).

Comparison of the different modes of employment reveals that our results generally show higher levels of employment within the creative industries (1.2 million people versus 1.1 million in 2001) and lower levels of embedded employment (645,000 versus 787,400). Because we can show where the embedded employment occurs across the broader economy,⁴¹ we feel that our calculations are the more robust.

Appendix 13 provides more details of the breakdown of the estimates for each of the creative segments for 1981, 1991 and 2001 and a comparison with the DCMS segment estimates.

Over the long period between 1981 and 2001 there have been substantial changes in the proportional breakdown of the segments, with growth in Advertising and Marketing as well as the Software, Computer Games and Electronic Publishing segment being particularly strong (Table 16). While some of these shifts inevitably result from classification changes, especially revised industry classifications, they also probably reflect genuine changes in the relative importance of these creative activities.

It is worth noting that each segments' share of creative employment has remained broadly flat since 2001 according to the LFS data.

The estimates in Appendix 13 reveal Architecture, Visual Arts and Design, Advertising and Marketing and Software, and Computer Games and Electronic Publishing to be the most embedded creative activities, with 52.9 per cent, 44.1 per cent and 35.9 per cent embedded shares in creative employment respectively. Music and Performing Arts are the least embedded, with an embedded employment share of just 15.6 per cent.

Table 16: Segment shares of overall creative employment for 1981 to 2006 and long-run growth rates of employment.

Source: Analysis by CCI of DCMS reports and custom census and LFS data tables from the Office for National Statistics

5.2 Creative Income Tridents: 2001 to 2006

Earnings in the creative sector are determined by combining the mean weekly income data specific to each combination of 'occupation within industry', as provided by the LFS survey on the basis of 'main job', with estimates of employment from the Creative Employment Trident (Table 17).⁴²

The LFS dataset provides a value for the average income for each year from 2002-2006 for each combination of creative occupation within each creative industry. Thus, we know that the average weekly earning for the 1,860 graphic designers working in businesses in the Book publishing industry is £268, which is less than the £929 a week the 1,378 Journalists and editors working in the same industry earn.

It is not straightforward to establish the annual earnings and mean incomes for census-based 2001 creative employment because census industry classifications are not at the same level of detail as those used in the LFS – so it is not possible directly to apply the value of average earnings for each specific industry/occupation from 2001 from the two data sources. Instead, we weight the means for each creative segment which we then apply to the segments from the census data.

The average weekly income figure for each occupation within industry is then multiplied by the number of people employed to generate the annual earnings for that combination (of occupations within industry). Dividing the 2001 annual earnings of each segment, for each mode, by the respective number of people employed produces a weighted mean annual income (for each segment, mode of employment and year).

The LFS-derived weighted means for 2001 are then applied to the Trident employment results of the disaggregated 2001 census data to generate census-employment-based annual earnings.

Table 17: The Creative Income Trident: total value of annual earnings generated through creative workforce salaries and wages for 2001 to 2006.

Source: Analysis by CCI of custom census and LFS data tables from the Office for National Statistics

It is worth noting that on these estimates the earnings from creative employment have risen to almost 10 per cent of the UK total workforce earnings in 2006 (Table 17). By way of comparison the DCMS Economic Estimates report shows that the Creative Economy's share of UK Gross Value Added is 7.3 per cent in 2005 (excluding Crafts and Design).

Parallels between annual earnings and Gross Value Added The methodologies and the datasets used to calculate the creative annual earnings (employment and average incomes) are in general different from those used for the determination of gross value added (GVA). Typically GVA estimates rely on input-output tables derived from surveys and often lag employment datasets in their release by several years. This raises the possibility that the earnings data may provide policymakers with a timely indicator of activity in the creative sectors.

Table 18: The mean annual income of the Creative Segments and the modes of employment for 2006.

Source: Analysis by CCI of custom LFS data tables from the Office for National Statistics

Comparing specialist, support and embedded average annual incomes The mean annual income in the UK in 2006, determined by data from the LFS for those in creative employment, was £28,770 (Table 18). This compares favourably with the mean for the UK workforce as a whole of £21,060. However, there is substantial variation between mean incomes across the segments and across the modes of employment.

According to these data, those working in Music and Performing Arts have the lowest mean income, at £21,880. Musicians working outside the creative industries appear to be particularly low paid, averaging £14,480. These results are similar to those found in the Australian data. Support workers in the software, computer games and electronic publishing segment are the highest earners with a mean income of £39,600.

5.3 Comparing creative incomes with UK average earnings

In line with the Australian Trident results, most of the modes of employment within the creative segments earn more than the overall UK average. The two main exceptions are the embedded modes within Music and Performing Arts and Publishing occupations which earn 31 per cent less and 12 per cent less, respectively, than the UK average (Table 19).

Table 19: Comparison of mean incomes of creatives with the UK mean in 2006.

Source: Analysis by CCI of custom LFS data tables from the Office for National Statistics

5.3.1 Summary of five-year growth rates

Bringing some of these data together, it is clear that over the period from 2001 to 2006, the growth rate of employment, earnings and mean income of the creative segment has been positive but slightly below that in the UK as a whole. Employment grew at 1 per cent per year on average compared with UK-wide growth of 1.2 per cent, while mean incomes grew 2.5 per cent per year versus UK-wide growth of 3.5 per cent (Table 20).

Four per cent growth per year on average in Music and Performing Arts employment combined with 3.7 per cent growth in mean income led to a growth in average earnings of 7.9 per cent, albeit from a low base. The Film, TV, Radio and Photography segment saw a shrinkage in employment of -1.5 per cent per year on average over this period, but an overall rise in mean incomes at 4.5 per cent.

The previous twenty-year period of high growth in employment in the Software, Computer Games & Electronic Publishing segment has slowed to just under 1 per cent per year on average with mean income close to the average growth rate for all creative segments at 2.3 per cent as a result.

At the other end of the spectrum the mean income for support workers in the Software, Computer Games & Electronic Publishing segment are 88 per cent higher than the national average, while Advertising specialists earn 70 per cent higher.

5.4 UK Whole-Economy Trident

5.4.1 The distribution of embedded creative employment in non-creative industries, 2001 to 2006

We have discussed how the detailed two-dimensional datasets used to calculate the Creative Trident allow additional types of analysis which are not possible using single-dimension, occupation-based or industry-based employment datasets.

It is particularly interesting to examine the distribution of embedded employment across the whole economy, either at the single-digit 'division' level or at the more detailed two-digit industry level. Table 21 also shows the Divisions in which the specialist creative industries are located; and within the creative industries, the breakdown between specialist and support employment.

The only creative employment within most Divisions is through embedded employment (the right-hand column), people working in creative occupations outside the specific creative industries. Outside Industry Divisions which include creative industries, embedded creative workers appear to be relatively more important in financial intermediation than in any other Industry Division. However, three Divisions (D, K, and O) also incorporate employment within the creative industries, in both specialist creative capacities and in embedded employment.

This Division K (Real estate, business activities) has the highest employment of specialist creatives, with 306,328 people, as well as the second-highest level of embedded employment (150,485). It is therefore worth examining in further detail (Table 22).

Division includes three creative industries classifications: '7220 Computer software consultancy', '74201 Architectural services' and '744 Advertising services', as well as 12 industries outside the 'creative' definition.

Unsurprisingly, 88 per cent of the creative occupations employed within 'Computer software consultancy' are also from that segment, with architecture, design and visual arts-related occupations accounting for 7 per cent. Similarly, 95 per cent of those employed in creative occupations within 'Architecture services' are architecture, design and visual arts-related.

However, the picture changes when the pattern of embedded employment is examined in non-creative industries in Division K, such as the very broad '741 Legal, accounting, business consultancy' which accounts for over 40,000 people in embedded creative roles, of which 37 per cent are in advertising and marketing functions, 33 per cent are in software development and 13 per cent are in architecture, design and visual arts roles.

5.4.2 The growth in creative employment within non-creative industries

As Table 23 illustrates, the Trident methodology provides a much stronger basis than single-dimension employment by industry or occupation tables for analysing and identifying patterns in employment. (See Appendix 14 for a series of tables showing the level of embedded employment for the industry subdivisions.)

Between 1981 and 2001 there was a substantial rise in the level of embedded employment across nearly all sectors of the economy (Table 23). The largest increase in the share of embedded creative employment was in Division J (Financial intermediation), up from 1.6 per cent in 1981 to 4.6 per cent in 2001, followed by Division I (Transport, storage and communication), where the embedded proportion increased from 0.6 per cent to 2.4 per cent. Only three divisions (K, N and O) showed a decline in the share of embedded creative employment over the period; the most significant of these was the fall from 6.1 per cent to 4.5 per cent in Division K.

Table 24 shows that for the shorter period 2001 to 2006 and using LFS data, there was no appreciable change in the share of embedded creatives in Division employment, except in Division L (Public administration), where the figure increased from 2.3 per cent to 3 per cent, and Division K, in which it appeared to rise slightly over the course of the five years.

Part 6: Applying the Trident methodology to Financial Services

One of the advantages of the Trident methodology is that it can also be applied to other industries and occupations allowing for inter-sector comparisons on a consistent basis. Using census data, this exercise has been undertaken for the UK financial services sector, permitting a point of comparison with creative industries and occupations.

Overall, on this measure, the 2001 Financial Trident is 8 per cent larger than the Creative Trident, with two million employed in the financial economy, 150,000 more people than in the Creative Trident. Over 1.2 million are employed in specialist financial services roles, almost 800,000 of whom are in financial roles outside the core financial services industry.

The picture of embeddedness this paints is similar to the Creative Trident, in that specialist occupations, whether creative or financial, account for a significant 63 per cent share of Trident employment, of which 34 per cent is embedded employment, in the case of the creative economy and 39 per cent in the case of financial services.

Interestingly total employment in the Financial Trident fell significantly between 1981 and 1991, from 2.9 million people to 1.8 million (Table 26), but recovered somewhat in 2001, growing to just over two million people, while the creative economy expanded throughout the period.

A substantial proportion of the drop in employment from 1981 to 1991 occurs within the occupation classification '400 Accounts and wages clerks, book-keepers, other financial clerks' which accounted for some 2.3 million people in 1981, whereas in 1991 the equivalent SOC '410 Accounts and wages clerks, book-keepers, other financial clerks' employed under 1.2 million people.

This change may reflect the productivity impact of computerisation on the financial management of businesses throughout the economy and the reduction in the employment of bank tellers in favour of ATMs. The difference in long-run trends between financial and creative employment raises the possibility that the extent to which technological change substitutes for skilled employment differs in the two sectors a topic for future research.

Table 27 breaks down finance employment across the wider economy, represented again by the Industry Divisions.

Some Divisions, such as Division K (Real estate, renting and business activities) employ very large numbers of people (approaching 300,000) in embedded finance roles; in this case accounting for 8.4 per cent of the Division's total employment. This reflects the employment of a range of financial consultants within '741 Legal, accounting, business consultancy'.

Of the 14 major industries, ⁴³ four have around 1.2 per cent of their workers in financial employment; this might plausibly represent the minimum level of financial specialists required for most business operations.

6.1 Parallels with the Creative Trident

The Industry Division Trident reveals further similarities between embedded employment in the financial and creative economies. In particular, while the levels are far from identical, there appears to be a moderate degree of correlation (51.2 per cent) between those sectors with higher finance embedded shares and those with higher creative embedded shares (Table 28).

One interpretation of this correlation is that most organisations in the knowledge and networked economy of the 21st century need a 'base level of service inputs – including creative and financial – to function efficiently, and very often they achieve this through direct employment rather than full or partial outsourcing to specialists. The differences between industries' levels of embedded employment could therefore reflect the frequency of use and strategic importance of these services and whether it is more effective for them to import them as embedded staff or as purchased services. Investigating this is a topic for future research.

Part 7: Conclusion: Main findings and implications

The purpose of this research report is to contribute to an improved evidence base for understanding how creative activities may fuel an innovative UK economy. The Creative Trident methodology offers a robust method for identifying all the components of creative employment: creative occupations within the creative industries, creative occupations in the rest of the economy, and non-creative (or support) occupations in the creative industries.

Census data have been used to construct the tables required to estimate the Creative Trident in the UK, supplemented by employment and earnings data from the annual Labour Force Survey. The three aspects of the Trident (creative employment, creative incomes and the whole-economy distribution) have been estimated for the three most recent UK census years (2001, 1991 and 1981), providing a view of changing activity in the creative economy.

The report arguably provides the most refined and robust picture of creative employment underpinning the creative economy in the UK, and relates its research findings to previous mapping work (particularly that of the DCMS).

7.1 Main findings

Finding 1: The creative economy accounts for over 7 per cent of UK employment, broadly consistent with the official estimates

The report shows that at the time of the 2001 census, creative employment accounted for 7.1 per cent of UK jobs, a very similar level to the DCMS Economic Estimates for the same period using different methodology, definition and datasets.

Finding 2: Creative employment has grown strongly over the long run, but has slowed down significantly in recent years

In addition, investigation of census data back to 1981 shows that UK creative employment grew at an annualised rate by 3.8 per cent between 1981 and 2001, compared with 0.8 per cent for the broader UK economy over the same period. However, growth seems to have slowed to 1.0 per cent per year between 2001 and 2006, just below the UK workforce annual rate of 1.2 per cent for the same period.

Finding 3: Average creative incomes are higher than in the economy as a whole, but have recently been growing more slowly

Creative incomes are on average approximately 37 per cent higher than in the UK economy as a whole in 2006, but had been growing at the slower rate of 2.5 per cent per annum since 2001, compared with 3.5 per cent for the total workforce.

Finding 4: There are more creatives working outside the creative industries than inside them

Compared with other economic activities, and consistent with other research commissioned by NESTA, creative employment occurs disproportionately outside the creative industries themselves, with 35 per cent of the total creative workforce (defined as specialist plus support plus embedded workers) embedded in non-creative sectors. A strong implication is that policymakers should recognise the limitations of using industry-based approaches alone to supporting the creative economy.

Finding 5: Creative workers are as embedded in the wider economy as financial services professionals

The 35 per cent of the total creative workforce that is embedded in non-creative sectors is similar to the 39 per cent of total UK financial services employment embedded in non-financial service industries at the time of the 2001 census.

Finding 6: Improvements in the availability of employment data are required to increase the robustness of the creative employment evidence base

Effective, evidence-based policymaking requires that greater attention be given to the frequency and coverage of official data sources, to data availability (to encourage exploratory research into employment characteristics) and to measurement issues (as emphasised by the DCMS's Creative Economy Programme), as well as to the policy implications of a focus on the role of creative activities across the economy as whole.

Our findings regarding the embedded nature of creative activities across the UK economy raise the possibility that the creative sector is significantly more involved in the wider innovation system than has been recognised to date. One possibility is that creative workers employed in the creative industries act as a conduit for knowledge and new ideas initiated in the creative industries. Investigating these channels is a priority for research, as it may have major implications for innovation policy, which has traditionally been focused on science and technology.

7.2 Implications: Improvement of data

Notwithstanding the groundbreaking quality of the DCMS Mapping Documents and the consistent value and timeliness of the DCMS Economic Estimates series, creative economy researchers in the UK face substantial challenges in having ready access to appropriate data. CCI's experience with obtaining, processing and analysing the datasets of the UK, NZ, Australia and the USA has highlighted a number of factors that can enhance or hinder a country's creative economy research capacity:

• data coverage and timeliness;
• levels of resolution and dimensions; and
• accessibility and consistency.

7.2.1 Data coverage and timeliness

Up-to-date population census and business register data provide the most accurate snapshot of the creative economy; but the ten-year interval between national censuses means that population data are unlikely to be current – a fundamental problem for researchers working in this field. Labour Force Survey data from a relatively small sample base, while timely, are no substitute for census data, as they do not facilitate multi-dimensional analysis; nor can they provide the accuracy at fine levels of classification that is required for most creative segment and creative sector analysis. Comparing the 2001 census data with the equivalent Labour Force Survey data shows a substantial difference in many of the relevant, most detailed industry classifications.

7.2.2 Data classifications: resolution and dimensions

Researchers in the UK have approximately half the industry classification resolution of their colleagues in the USA, Australia and New Zealand.

In the occupation dimension, UK researchers fare even worse, with less than half the level of classification resolution available in the USA, and less than a third of the Australian and New Zealand levels.

These classification constraints inevitably flow through to the level of detail within the datasets available to researchers, and limit their ability to produce more robust estimates. Without considering the number of additional records involved in providing the mean income data, researchers in the UK have, at best, access to a third of the employment data records available to researchers in Australia and the USA. ⁴⁴ The preferred Australian dataset used for Creative Trident analysis at the national level has three dimensions – the count of employment within 16 income bands, 716 industries and 1,334 occupations. After removing 12,954,480 null records, this dataset contained 417,536 records of detailed data. There is no equivalent dataset available in the UK.

7.2.3 Dimensionality

Full resolution, multi-dimensional employment datasets need to be available for researchers to process, link with their other datasets and to explore in the course of their research. A single dimension (e.g. industry only) or summary cross-tabulated two-dimensional data does not support sufficiently robust analysis.

7.2.4 Accessibility and consistency

Research data, such as employment, should be as 'analysis friendly' as possible, especially so for official statistics. The more difficult the data are to obtain, the harder they will be to process for analysis.

7.2.5 Implications of the UK Creative Trident mapping for data quality

In applying a methodology first developed for Australian data, we have been able to take a fresh perspective on data quality as it relates to the UK's creative employment. The study has highlighted an opportunity for the DCMS to work with the Office for National Statistics to improve relevant datasets over the medium to long term. Priorities include:

Classifications

• Increase the resolution of UK occupation and industry classifications by extending the number of digits while maintaining compliance with EU and international requirements.
• Reprocess and recode the UK 2001 census data at consistently higher resolutions of occupation (six-digit) and industry (five-digit) classifications. At the very least, recode into finer classifications all returns in which a single industry classification has over 50,000 people employed.
• Implement a policy to release research data at a common level of classification and detail that is, not mixing four- and five-digit detail along with subtotal and total data within a given dimension.

Data collection

• Increase the sample size of the LFS substantially to increase its accuracy and to allow for the release of full-resolution 'occupation within industry' employment data.
• Use the LFS to collect the information that is deemed to be equivalent to personal income details for the self-employed.
• Move to conduct either an employment/population census or a significantly expanded Labour Force Survey every five years, using the full resolution of employment classifications and collecting personal income information.

Fostering a 'culture of metrics'

The shift to evidence-based policy would benefit from the development of an even stronger metrics culture. This would be facilitated by:

• Making the source employment and industry performance data available online for researchers to download. While reports provide analysis and stimulate discussions, they do not remove the barriers faced by other researchers attempting to replicate, extend or challenge the findings. At the very least, the source data for all official reports such as the DCMS Economic Estimates should be available for others to build on and challenge (all the data used in this analysis will be posted on NESTA's website).
• Making multi-dimensional census and LFS employment data easier to access and analyse via direct download and through an online analytical service/statistics repository.

Our findings regarding the embedding of creative activities across the economy raise the possibility of cross-industry linkages and 'technology transfer' due to creative workers, and mean that the creative sector may be significantly more involved in the innovation system of national and regional economies than has been recognised before. This may have important implications for innovation policy which has traditionally been exclusively associated with the science- and technology-based industries.

7.3 Creative industries articulated into innovation policy

This report supports a shift in focus for policy from creative outputs (the creative industries as a specific sector) to creative occupations as inputs into the whole economy, and creative outputs as intermediate inputs into other sectors. As such, it complements other research commissioned by NESTA as part of its Arts and Innovation research programme. This idea of creativity as an economic 'enabler' arguably has parallels with the way financial activities and information and communication technologies have been shown to be broad enablers of economic growth in the past. This may facilitate a stronger focus on innovation systems which support the development of the creative economy.

Appendix 1: Selected creative occupations and industries: Census and LFS

Footnotes

Appendix 2: The alignment between the detailed census and LFS employment data

Figure 2 plots the frequency of the discrepancy between the employment data used from the 2001 census and the LFS. The LFS four-digit industry classifications are aggregated to the closest detailed census classification which may be two- or three-digit.

Figure 2 reveals a substantial difference between two data sources in the numbers of people employed in most of the relevant industry and occupation classifications - in some cases by as much as 20 per cent to 30 per cent, despite the total employment figure showing only a 1 per cent discrepancy.

Figure 2: Frequency of discrepancy between census and LFS employment - industry classifications.

Source: Office for National Statistics 2001 Census of Households and 2001 LFS

Similarly, when employment data are compared on the basis of occupations, there is a significant variation at the four-digit classification level, with over 30 per cent of classifications having a difference exceeding 20 per cent (Figure 3).

Reliably measuring creative employment requires the summing-up of many detailed classifications; and since the LFS data appear to be so much at variance with census data, it would seem likely that the LFS data in its current form is not as accurate as would be hoped, for use as the basis for absolute measurement of creative employment, nor for tracking the subtler shifts over time.

Figure 3: Frequency of discrepancy between census and LFS employment - occupation classifications.

Source: Office for National Statistics 2001 Census of Households and 2001 LFS

Appendix 3: The impact of excluding second jobs from LFS based analyses

Adding second jobs to the employment basis for the calculation of the Creative Employment Trident does little to increase the relative size of the creative economy and at the same time makes interpretation of the results more difficult - it shifts the basis for all calculations from the number of people employed to the number of jobs.

What follows are experimental estimates to determine the impact of second jobs and the difference between a creative workforce Trident and a creative jobs Trident. They should not be interpetated as providing an accurate estimate of the number of creative jobs.

Using main and second job data from the 2006 LFS dataset shows that the creative share would grow from 7.8 per cent of total employment (by definition main job) to a 7.9 per cent share of total jobs.

Overall, the incidence of second jobs in the creative workforce is 5.5 per cent compared with 3.7 per cent for the total UK workforce (Table 36).

Table 36: The number of people with second jobs and main jobs in creative occupations for 2006.

Source: CCI analysis of custom LFS data from the Office for National Statistics

Table 37: The number of people with second jobs and main jobs in creative segments for 2006.

Source: CCI analysis of custom LFS data from the Office for National Statistics

The high ratio of second jobs in Music & Performing Arts occupations (39.5 per cent) is also reflected within Music & Performing Arts industries.

Table 38: The ratio of the number of people with second jobs to those with main jobs in creative occupations and industries for 2006.

Source: CCI analysis of custom LFS data from the Office for National Statistics

The following table breaks down first and second creative jobs by occupation and industry.

The table reveals the extent of the diversity of employment: for example the Music and Performing Arts Industry is a significant employer, on both a main and second job basis, of design, film and publishing-related professionals.

Table 39: The number of people employed in main and second creative jobs by occupation and industry segments for 2006.

Source: CCI analysis of custom LFS data from the Office for National Statistics (some cells have been subject to suppression).

Conversely, the Software, Computer Games & Electronic Publishing industry employs relatively few individuals on a second job basis. All these estimates are of course subject to possible finite sample biases insofar as the LFS is only a sample-based survey.

Appendix 4: The Disaggregating of 7420 Architecture and Engineering Services in the LFS dataset

The disaggregation of '7420 Architecture and Engineering Services' into '7420/1 Architecture Services' and into an artificial industry '7420/9 Engineering Services' was not trivial, as adjustments had to be made at the detailed level and to each dimension's subtotals to maintain the integrity of the datasets.

In summary it involved taking the employment count within '7420 Architecture and Engineering Services' for each creative occupation (and subtotals) for each year of the LFS dataset.

Not all of the employment in creative occupations is allocated to the disaggregated creative industry '7420/1 Architectural Services'. Some is allocated to '7420/9 Engineering Services' to reflect the fact that some engineering service firms employ graphic artists, designers, programmers and marketeers (Table 40).

Table 40: The allocation of creative occupations across two disaggregated industries.

Source: CCI analysis of custom LFS data from the Office for National Statistics

While this procedure results in no change to total employment in any of the years, it does mean that there is a shift between the modes: in any one year, specialist employment decreases by approximately 10,000 with a balancing increase in embedded employment. The level of support employment falls by around 150,000 each year as this amount is shifted to non-creative employment (Table 41).

Table 41: Disaggregating 7420 Architecture and Engineering Services from 2001 to 2006.

Source: CCI analysis of custom LFS data from the Office for National Statistics

Appendix 5: The impact of classifications on creative employment measures

In 2006, the Australian Bureau of Statistics adopted new classification schemes for occupations and industries. The direct impact of a more modern classification scheme on creative employment can now be better understood as the 2006 Australian census data was coded twice - under the previous scheme and under the new scheme.

An examination of the number of classifications actually used in the census reveals a significant increase in the number of occupation classifications.

An increase in the number of available classifications increases the ability to discriminate, but also generally results in an increase in the level of a segment's measurable employment. The main benefactor of this classification change is the Software and Digital Content segment, which expanded from two to seven industry classifications and from four to 19 occupation classifications. Unsurprisingly, its employment increased 16 per cent in total.

Table 42: The number of classifications available in the 2006 Australian census for measuring the creative segments.

Source: CCI analysis

Table 43: The impact of the change in classifications on the measurement of creative employment in Australia.

Source: Analysis by CCI of Australian Bureau of Statistics 2006 Census of Population and Housing

Appendix 6: Top ten occupations employed in each creative industry

The following table is an example of the range of tables used to determine the relevance of industry classifications to the Creative Trident. The table is generated from the two-dimensional 2001 census matrix containing the number of people employed within each occupation in each industry.

Table 44: The top ten occupations within each candidate creative industry using 2001 census classifications and data.

221 Publishing

362 Manufacture of jewellery and related articles

7220 Computer Software consultancy (After disaggregation from 72 Computer Activities)

74201 Architectural activities and related technical consultancy

744 Advertising

920 Recreational, cultural and sporting activities n.o.s.

921 Motion picture and video activities

Table 34: Core creative occupations selected from those used in the 1991 census.

Table 35: Core creative occupations selected from those used in the 1981 census.

Source: Office for National Statistics 2001 Census of Households

Appendix 7: Top ten industries of employment of creative occupations

The following table is used to determine the relevance of occupation classifications to the Creative Trident. The table was generated from the two dimensional 2001 Census matrix containing the count of the people employed within each occupation within each industry. This level of analysis cannot be conducted with the same level of confidence on LFS datasets because the matrices contain too many suppressed cells.

Table 45: The top ten industries for each candidate creative occupation using 2001 census classifications and data.

Source: Office for National Statistics 2001 Census of Households

Two occupations included within the DCMS definitions were initially considered by CCI to fit the definition of creative occupations, and LFS tables were ordered on this basis. However, on examining the census matrices of the industries in which these occupations were employed, it became clear that they did not meet one of the usual characteristics – typically, creative industries account for at least 25 per cent of the employment of a creative occupation.

At best, only 18 per cent of those employed in '2126 Design and development engineers' are employed in the creative classification '742 Architectural activities and related technical consultancy', but the figure is in fact far lower, because that industry itself is very broadly defined and includes a substantial proportion of technical and engineering occupations. After disaggregation, the creative industry share of employment would be less than 10 per cent. The second occupation, '5499 Hand craft occupations n.e.c.', belies its name and is substantially involved with car repairs and manufacture. It does not correlate with the arts and crafts occupations that its name suggests.

Table 46: The top ten industries for the '2126 Design and development engineers' occupation using 2001 census classifications and data.

Source: Office for National Statistics 2001 Census of Households

Table 47: The top ten industries for the '5499 Hand Craft occupations not elsewhere classified' occupation using 2001 census classifications and data.

Source: Office for National Statistics 2001 Census of Households

Appendix 8: Results of applying the Trident methodology to Australian datasets

The Trident methodology has now been applied to Australian Census data from the years 1976, 1981, 1996, 2001 and the 2006 dataset that was released in October 2007.

The results of the studies have revealed the patterns of growth, the shifts between segments and the extent of embedded employment in the wider economy with much greater clarity and consistency.

Over the period there was a sustained increase in embedded employment's share of total creative employment, from 31 per cent in 1996 to 35 per cent in 2006, with some segments such as Film, TV and Radio showing significantly higher increases.

Table 48: Total Australian creative employment 1996, 2001 and 2006 and the 10-year annualised growth rate.

Source: Analysis by CCI of Australian Bureau of Statistics 1996, 2001, 2006 Census of Population and Housing

Table 49: Embedded employment's share of creative employment by segment between 1996, 2001 and 2006 and the 10-year annualised growth rate.

Source: Analysis by CCI of Australian Bureau of Statistics 1996, 2001, 2006 Census of Population and Housing

Income patterns

The Australian census datasets are available at much higher resolution than the equivalent UK datasets and also include, depending on purpose, either the mean income or the level of employment in 16 bands of income. Much more detailed analyses of income distribution can therefore be conducted with confidence.

The following examples illustrate what analysis can be done with such data were it to be available in the UK.

The patterns of income distribution within the Australian segments 2001

There were 150,000 people employed (34 per cent) across the six creative segments who were earning less than $600 (equivalent to £258) per week in 2001 dollars. Of these, 77,000 earned less than $400 a week. However this represents a much lower proportion of total creative workforce employment (18 per cent) than the population as a whole (27 per cent).

The patterns of distribution of weekly incomes vary substantially across the segments. This can be seen in Figure 4 where the lines represent a particular weekly income band's share of the total number of people employed in that segment. The shaded area of the charts also shows the distribution of income for all people employed in Australia.

Compare the distribution pattern of the Music segment (the green line) which has a fatter 'tail' (more people than normal earning less money) and a smaller 'head' (fewer people earning higher money) to that of two other segments - Software and Film – which exhibit tiny tails and fat heads (a steady slope up to the right indicating that a higher proportion of their total employment is at the highest income levels).

(Bar chart showing employment distribution within weekly income bands for various creative segments compared with Australian employment.
X-axis: Weekly income bands (e.g., $1-$39, $40-$79, ... $1,500 or more)
Y-axis: Percentage (0% to 25%)
Multiple lines representing different segments: All Employed, Design, Advertising and Marketing, Film, TV and Radio, Music and Performing Arts, Publishing, Software and Interactive Content, Creative Industries Total.
The chart shows variations and peaks in certain income bands for different segments, generally showing higher income for creative industries compared to "All Employed" in the higher bands.)

Source: Analysis by CCI of custom ABS 2001 Census tables

Australian mean annual personal incomes, 2001

Generally, those in the creative workforce who were employed within creative occupations had a higher annual mean income than those employed within the same Industry Division.

Creatives embedded in Communication Services and Finance and Insurance earned 44 per cent and 42 per cent respectively more than their Division average, while those working in Manufacturing earned 12 per cent more than the average across the Division (Table 50). Workers in the Writing, Publishing and Print Media segment (which forms part of Manufacturing) were 7 per cent above the Division average.

Source: Analysis by CCI of custom ABS 2001 Census tables

Notably, the creatives Embedded within the Education division had annual salary averages 18 per cent below the Division average.

Appendix 9: Self-employment rates within the creative occupations and industries

Self-employment of individuals with creative occupations

The proposition that self-employment is the norm for the creative workforce is not borne out by an examination of the facts. While the rate of self-employment for creative occupations in the UK is roughly twice that of the general workforce it still represents only 28 per cent of the employment of those in creative occupations (Figure 5).⁴⁵

(Horizontal bar chart showing self-employment rates by creative occupation segment.
X-axis: Percentage (0% to 100%)
Y-axis: Occupation segments (e.g., Advertising and Marketing, Architecture, Design and Visual Arts, Film, TV, Radio and Photography, Music and Performing Arts, Publishing, Software, Computer Games and Electronic Publishing, All Creative Occupations, All Professionals, All Associate Professional, All Employed)
Key rates:
- Advertising and Marketing: 13%
- Architecture, Design and Visual Arts: 33%
- Film, TV, Radio and Photography: 47%
- Music and Performing Arts: 82%
- Publishing: 28%
- Software, Computer Games and Electronic Publishing: 15%
- All Creative Occupations: 28%
- All Professionals: 14%
- All Associate Professional: 14%
- All Employed: 13%
)

Source: CCI analysis of custom LFS data 2006

As Figure 6 shows, there is great diversity across segments in the self-employment rate, depending on occupation, individual preferences, age and possibly skill level: some occupations naturally have substantially higher numbers of self-employed people such as '3411 Artists' with 92 per cent self-employment and '3414 Dancers & choreographers' at 100 per cent. However others are perhaps lower than average, for example '3412 Authors writers' with 62 per cent and '3421 Graphic designers' with 32 per cent self-employment.

Self-employment within creative industries

The patterns are similar when looked at from the dimension of creative industries. Table 51 shows that across all creative industries, the self-employment rate is 27 per cent, compared with the whole-economy rate of 13 per cent. The highest rate is in Music and Performing Arts (64 per cent), followed by Film, TV, Radio and Photography (37 per cent); Software, Computer Games & Electronic Publishing has the lowest rate (17 per cent).

(Detailed horizontal bar chart showing self-employment rates for disaggregated creative occupation segments.
X-axis: Percentage (0% to 100%)
Y-axis: Specific occupations within broader segments (e.g., 1134 Advertising and public relations managers, 3411 Artists, 3414 Dancers and choreographers, 3412 Authors, writers, etc.)
Rates vary widely, highlighting the diversity mentioned in the text.)

Source: CCI analysis of custom LFS data 2006

Of greater importance perhaps to research on the sustainability of the creative economy is the proportion of sole-practitioner businesses within segments and the pattern of distribution of income.

Australian census data for 2006 show that, on average, those who own a creative business employing up to 19 staff earn a 50 per cent higher mean income than sole-practitioners in the same industry, while owners of a creative business employing 20 or more staff earn 73 per cent more than the smaller business owners.

It would seem that creatives who lack the business acumen to know how to grow their practice to the point that it can support a number of employees are relegated to the lower rungs of the earnings ladder. With some notable exceptions, creative excellence alone in one's chosen craft is no guarantee of a reasonable income.

Source: Analysis by CCI of Australian Bureau of Statistics 1996, 2001, 2006 Census of Population and Housing

Appendix 10: Mapping creative industry-relevant census SIC to LFS SIC codes

The following table correlates the industry classifications and the employment data used for the 2001 census and the LFS. The LFS four-digit industry classifications are aggregated to align with the closest detailed census classification, which may be at two digits ('72 Computers') or three. The LFS employment figures are used to establish the relevant proportion of the employment at the correlating census classification. As previously discussed, disaggregation is not synonymous with simple scaling or applying a proportional calculation.

Appendix 11: Coding Lists

An example of the job titles that are coded within standard occupation classifications

Every industry or occupation classification system is constrained in its ability to capture the variety of names given to people's jobs and primary activities in the real world. Coding lists are used to provide consistency in references to similar functions or business activities. Every formally identified occupation classification may go under ten, 40 or in some cases even 100 other names which are in common usage. These terms are often synonyms, but some classifications are treated more as a 'grab bag' of assorted activities which may have only a cursory correlation, such as being mainly self-employed artists, musicians, writers. The following is a small extract from the SOC2000 coding list.

Appendix 12: Census 2001 – Disaggregating the data records of two industry classifications

Disaggregating '72 Computer and related activities'

The census employment for the industry classification '72 Computer and related activities' is 540,290, which is comparable to the figure under LFS for 2001 (531,960) when five 4-digit SIC industries are aggregated to two digits.

The four-digit industry classification relevant to our methodology is '7220 Computer Software consultancy' which accounts for 69 per cent of the LFS total for '72 Computer and related activities'. This factor is then applied to the census total to derive a gross figure for 7220 Computer Software consultancy (Table 54).

The target specialist to support an employment ratio of 53 per cent: 47 per cent (Table 55) for the disaggregated industry is determined with reference to the proportions within the custom LFS occupation within industries sub-totalled matrix acquired from the ONS.

These factors are then applied in a matrix to determine the new target figures for the 7220 category, relegating the balance to '7260 Other computer activities'.

Source: CCI analysis of custom LFS data 2001 and Household Census Data 2001

Source: CCI analysis of custom LFS data 2001 and Household Census Data 2001

A subset of the 2001 census employment database of occupations by industries is then established, containing 350 records – each record consisting of the number of people employed within one of the 350 occupations that also have employment within the industry classification '72 Computer and related activities'.

The details of employment within each occupation are then proportionately assigned to either '7220 Computer Software consultancy' or '7260 Other computer activities' on the basis of the relevance of the occupation to the software development and publishing activity, with a proportion of this employment being allocated to the remainder industry (reflecting the dispersal in the real world).

The result is 698 records with the allocation of the employment in total matching the LFS split ratio and the specialist support ratio for '7220′.

The existing 350 records for '72 Computer and related activities' are then deleted from the 2001 employment database, and the 698 disaggregated records imported.

Disaggregating '742 Architectural and engineering'

Disaggregating '742 Architectural and engineering' into the architecture-relevant activities from the engineering activities was more difficult, as no UK employment data are available for estimating the differential proportions on the basis of industry. So instead we examined the range and proportion of occupations employed within the industry in the 2001 census.

The occupations are grouped into 'Architecture and Design occupations', 'Other technical services occupations such as engineers' and 'General office, administration and support occupations' (Table 56).

Source: Office for National Statistics 2001 Census

Source: Office for National Statistics 2001 Census

The ratio between Architecture and Design Occupations and the Other Technical Services Occupations was calculated to be 43 per cent to 57 per cent and this is applied as one factor to split employment into two classifications: '74201 Architecture Services' and a custom classification we created, '74209 Other Technical Services'.

The same process as is used in disaggregating '72 Computer and related activities' is applied to the master employment database, and the initial 350 records for '742′ became 430 records across the two new industries, while maintaining the correct totals for creative occupations and 'other' or support occupations.

Appendix 13: Detailed tables and comparisons with DCMS Economic Estimates

This section shows a breakdown of specialist and embedded employment by creative segment (where creative segments comprise multiple three- or four-digit sub-divisional industries, sometimes from across different Divisions). As shown below, this allows analysis of the variations across segments in the proportion of specialist, support and embedded employment.

Total creative employment grew by 1991 to just over 1.1 million people.

Source: Analysis by CCI of custom 1981 Census data tables from the Office for National Statistics

Source: Analysis by CCI of custom 1991 Census data tables from the Office for National Statistics

As Table 60 shows, employment in 2001 had grown to 1.9 million people.

Source: Analysis by CCI of custom 2001 Census data tables from the Office for National Statistics

While the total level of employment between 2001 to 2006 changed by less than 100,000, Table 61 shows there were quite major shifts in the proportions between specialist and support staff – with an apparent increase in the number of specialist staff and decrease in support staff and to a lesser extent between specialist and and embedded employment. We cannot rule out however that some degree of this could result from the lower level of detail of the 2001 Census Industry classificiations.

Source: Analysis by CCI of custom 2006 LFS data tables from the Office for National Statistics

The following figures show the degree of correlation between CCI's estimates employment within the segments and those in the DCMS Economic Estimates.

(Stacked bar chart showing total employment (up to 2,000,000) over the years 1981, 1991, 2001-2006.
For each year, there are two bars: 'Census' and 'Economic Estimates'.
Each bar is segmented by creative industry: Advertising and Marketing, Architecture, Visual Arts and Design, Film, TV, Radio and Photography, Music and Performing Arts, Publishing, Software, Computer Games and Electronic Publishing.
The chart visually compares the employment figures derived from CCI's analysis of Census/LFS data against DCMS Economic Estimates for each segment over time.)

Source: Analysis by CCI of DCMS reports, and custom Census and LFS data tables from the Office for National Statistics CCI

Each segment's share of total creative employment is shown in Figure 8, which also illustrates the long-run growth in the Software, Computer Games and Electronic Publishing segment and the declining shares of the Publishing and Architecture, Visual Arts and Design segments.

(Stacked bar chart showing the percentage share of segment employment (0% to 100%) over the years 1981, 1991, 2001-2006.
For each year, there are two bars: 'Census' and 'Economic Estimates'.
Each bar is segmented by creative industry: Advertising and Marketing, Architecture, Visual Arts and Design, Film, TV, Radio and Photography, Music and Performing Arts, Publishing, Software, Computer Games and Electronic Publishing.
The chart visually compares the proportional contribution of each creative segment to total employment, highlighting changes over time between CCI's data and DCMS estimates.)

Source: Analysis by CCI of DCMS reports and custom Census and LFS data tables from the Office for National Statistics

Figure 9 shows how the average annual earnings of most specialist, support and embedded staff working in the creative economy are higher than those of the national average. The poorest performing areas are embedded professionals in Music and Performing Arts,⁴⁶ and Publishing, but even there the average income of music has grown strongly and recovered some of the gap with the national average.

(Line graph showing average annual incomes (£0 to £40,000) for different categories (Specialist, Support, Embedded, Creative Employment, UK workforce) across various creative segments (Software, Computer Games & Electronic Publishing; Advertising and Marketing; Total of all segments; Film, TV, Radio and Photography; Architecture, Visual Arts and Design; Publishing; Music and Performing Arts) from 2001 to 2006.
The graph displays multiple lines, indicating trends in income for different roles within each segment, and compares them to the 'UK workforce' average. Generally, creative employment categories show higher incomes than the UK workforce average, with some fluctuations.)

Source: Analysis by CCI of DCMS reports and custom Census and LFS data tables from the Office for National Statistics

Appendix 14: Embeddedness by Two-digit Industries

Table 62: The 25 sub-divisions with the highest proportion of embedded employment 2001 UK census data.

Source: Analysis by CCI of custom Census 2001 table from the Office for National Statistics

Table 63: Growth in embedded creative employment across the Industry sub-sectors.