Skip to main content

Where Is Creativity? Data and Methodology to Measure CCIs Across EU Regions

  • 81 Accesses

Part of the Contributions to Regional Science book series (CRR)


The quality of the analysis of CCIs locations depends on the quality of data employed. Official statistics miss information that is finely disaggregated, both at the industrial and at the geographical level. The aim of this chapter is to present a novel database built for this work, starting from Orbis data. The most relevant aspect is to measure two determinants of CCIs: creative employment, considered a proper indication of the amount of creatives in a given location, and the degree of innovativeness of these industries in space. The richness of the database created allows maps on the geography of CCIs to be produced according to their different degrees of innovation intensity.

This is a preview of subscription content, access via your institution.

Buying options

USD   29.95
Price excludes VAT (USA)
  • Available as PDF
  • Read on any device
  • Instant download
  • Own it forever
USD   99.00
Price excludes VAT (USA)
  • Available as EPUB and PDF
  • Read on any device
  • Instant download
  • Own it forever
Hardcover Book
USD   129.99
Price excludes VAT (USA)
  • Durable hardcover edition
  • Dispatched in 3 to 5 business days
  • Free shipping worldwide - see info

Tax calculation will be finalised at checkout

Purchases are for personal use only

Learn about institutional subscriptions


  1. 1.

    ATECO is the Italian classification of economic activities (ATtività ECOnomiche). The 2002 version is based on NACE Rev. 1.1.

  2. 2.

    The correspondence table (tavola di raccordo tra ATECO 2007 e ATECO 2002) is available through the ISTAT website.

  3. 3.

    The database allows mitigation of the possible trade-off between industrial and spatial disaggregation that may emerge in studies like this one. In fact, because of privacy issues, it is usual for firms not to disclose at the same time details on industrial and spatial details. In that case, it would be possible to identify firms that produce a very specific type of good.

  4. 4.

    Official news on the acquisition of BvD available at Moody’s Corporation—Press Releases.

  5. 5.

    To avoid misconducts, the access to the Historical Orbis platform was allowed only to a dedicated PC at the Technology Transfer Office of the Politecnico di Milano. I thank Massimo Barbieri for his support during this process.

  6. 6.

    In most of the cases, each record found in Orbis refers to a specific establishment, even if it belongs to a larger company. For instance, this is the case of national-level companies belonging to a larger group (e.g. Adidas has national branches, each of them representing a single entity). Dropping C2 records from the sample has exactly the aim of preventing the data from containing both the holding company, embedding all employees from all branches, and branches themselves. Therefore, through this methodology the database contains all the different establishments of a group if the information is separated and available; otherwise, if only the data for the headquarter is available, this is considered alone.

  7. 7.

    Available at

  8. 8.

    The international classifications for patents (IPC) and trademarks (NCL) can be consulted through the WIPO.

  9. 9.

    These variables are in an industry-region scale. The average indicated here refers to the average productivity measured across industries, for each region.

  10. 10.

    The subdivision of manufacturing codes according to high-tech propensity follows the criteria provided by Eurostat: Indicators on High-tech industry and Knowledge—intensive services.

  11. 11.

    The ownership of works of art, literature, music, multimedia and other protectable works in general resides in their creators (see the EUIPO website for further information).

  12. 12.

    FIGARO—Experimental statistics—Eurostat. The FIGARO tables were born to analyse the socio-economic and environmental effects of globalisation also through global value chain relationships.

  13. 13.

    The logic can be replicated using the median and not the mean, but minor changes would apply with no relevant theoretical implications.

  14. 14.

    The share is calculated using the overall regional employment as denominator. The source can be either Eurostat or ARDECO. The spatial distribution does not present major changes in either of the two cases.

  15. 15.

    For an overview of the company, refer to

  16. 16.

    Cf., accessed 31/01/2022.

  17. 17.

    Cf., accessed 31/01/2022.

  18. 18.

    LISA are constructed by local Moran’s Is using the toolkit present in GeoDa (Anselin et al. 2002, 2010).

  19. 19.

    Usually, international bodies dealing with IPRs prefer an industrial classification of sectors according to the intensity in producing patents, trademarks, or copyrights (EPO and EUIPO 2016; ESA and USPTO 2016; USPTO 2012).

  20. 20.

    Technical aspects related to ANOVA are presented in the Appendix 5.2 to this chapter.

  21. 21.

    Only significant correlations are shown (p-value < 0.05).

  22. 22.

    Post-hoc tests are conducted after the fact, i.e. after a significant ANOVA, and they are used to evaluate among which groups the significant differences exist.

  23. 23.

    In Stata, the pwmc command allows to perform multiple comparisons, using methodologies to correct for unequal variances.


Download references

Author information

Authors and Affiliations


Corresponding author

Correspondence to Roberto Dellisanti .


Appendix 5.1

See Table 5.12.

Table 5.12 Conversion of Santagata’s codes according to NACE Rev. 2 classification

Appendix 5.2

The first of the two ANOVA tests presented in this chapter is a multiple comparison of means across quintiles of Inventive and Replicative CCIs. Table 5.13 presents the results of both the F-test of equality of group means and the Bartlett’s test for equal variances among the groups. Both tests are important because they guide the following steps. In fact, accepting the null hypothesis of the F-test means that the model is not able to detect any differences among the quintiles. Moreover, the response of the Bartlett’s test is even more important. In case of unequal variances, a correction method should be applied in the post-hoc test for the differences.Footnote 22 In case of unequal variances, Tamhane’s T2 post-hoc tests for pairwise comparisons of means are used. Indeed, in case of unequal variances among groups, classical adjustments (e.g. Bonferroni, Scheffé) are not suitable options and it is necessary to correct for this (Tamhane 1979).Footnote 23

Table 5.13 Tamhane’s T2 post-hoc test of pairwise comparisons among quintiles

Instead, the second analysis was performed to compare Inventive and Replicative by the size of these clusters (i.e. each quintile of the distribution). A t-test is conducted as it involves only two groups. More specifically, a Levene’s test is used to assess the equality of variances for a variable calculated for two or more groups (Levene 1960). When the null is rejected, a more generalised version of the t-test should be performed. The most used and applied in this work is Welch’s correction (Welch 1947). Results of this analysis are presented in Table 5.10.

Looking at the results, there emerge substantial differences among the localisation patterns of different CCIs. First, although both low Inventive and low Replicative regions are largely located in the Eastern part of the continent, for high quintiles this is true only for Replicative. Furthermore, it is possible to state that Replicative CCIs largely prefer Eastern regions. Second, considering population levels and GDP, the comments are specular. Indeed, for both variables, it is not possible to detect any difference among quintiles of Replicative CCIs while for Inventive differences exist and are relevant. The more Inventive employment is hosted in a region the higher is the population and the economic size of the region itself. This indicates that Inventive CCIs may have a strict preference for large urban areas, where generally new ideas emerge more easily and, reversed, only strongly creative CCIs may bear the cost of urban locations. This reasoning is also supported by the results for the dummy Metropolitan area that follows the same scheme. Third, the regional knowledge environment, proxied by the patents per capita and the education levels, goes in the direction of the previous results. Indeed, these variables go hand in hand with concentration of Inventive CCIs, while understanding the correlation with Replicative employment is much less straightforward. In other words, innovations call for innovations and Replicative CCIs are looking for something else. Summing up the ANOVA and post-hoc results, it is possible to say that Inventive CCIs follow a well-established scheme in regional sciences: as they are driven by innovations, they tend to cluster where there exist the conditions for the flourishing and the valorisation of new knowledge. These places are mostly Western cities and areas with highly developed knowledge environments. Instead, Replicative CCIs are much more difficult to pigeonhole as their clustering is more chaotic and less explained by classical variables of regional economic literature. The idea is that, if not innovative, these activities may cluster according to static local conditions that improve their efficiency. On the contrary, Inventive CCIs are expected to follow a dynamic logic of clustering, looking for factors that may stimulate their innovative capacity.

Rights and permissions

Reprints and Permissions

Copyright information

© 2023 The Author(s), under exclusive license to Springer Nature Switzerland AG

About this chapter

Check for updates. Verify currency and authenticity via CrossMark

Cite this chapter

Dellisanti, R. (2023). Where Is Creativity? Data and Methodology to Measure CCIs Across EU Regions. In: Cultural and Creative Industries and Regional Development. Contributions to Regional Science. Springer, Cham.

Download citation