Reasons for Clustering of Creative Industries in Italy and Spain

Abstract

Creative industries and creative employment tend to concentrate around medium and large cities, forming creative local systems. We follow a multidisciplinary approach, based on cultural and creative economics, evolutionary geography and urban economics, in order to analyse the forces behind the clustering of employment in creative industries in a comparative analysis of Italy and Spain. The results show different patterns of clustering of creative employment in both countries. The historical and cultural endowments, the average size of creative industries, the size of the place, the productive diversity, and the concentration of human capital and creative class have been determined to be common factors leading to a concentration of creative firms and creative employment in both countries.

Notes

Florida's theory has been criticized by some authors. Despite these criticisms, the contribution of the “creative class” theory is recognized as redirecting the attention from the firm to the creative class, and so to the qualified human capital (Asheim, 2009).

We refer to Lazzeretti et al. (2008) for a wider discussion on the use of the LQ as an indicator of specialization-concentration for creative industries. The choice is also pragmatic: we have used other indicators (simple concentration coefficient, Florida mixed indicator, etc.) and other versions of the LQ, but the basic LQ continues to be the most easily interpretable and produces more solid results.

Using firms as the basic unit of measurement of concentration does not produce very different results in the maps of creative local production systems or in econometric regressions, which have been interpreted as a sign of robustness. In both cases (employment and firms), the data set includes the complete population in both countries: wage-earners, self-employment and freelancers, special regimes, large firms, small firms, micro-firms, etc.

Lorenzen and Andersen (2009) depart from a similar logic to discuss the spatial hierarchies of the creative class.

Other tests have been performed, departing from a more general translog function, although we arrive at the same log-linear equation.

We have also considered the number of museums localised in the local labour systems registered by the Ministry of Culture in the two countries as a proxy for the cultural heritage and a dummy representing the UNESCO World Heritage List, although results have not been significant. The scarcity of statistics about heritage and cultural endowments or expenditures at a local level limits this part of the analysis.

The value of the entropy indicator increases the more diversified the creative profile of an LPS is, where is the aggregation from three-digit to two-digit sectors of the share of each industry's employment of the total employment , and .

Florida (2002) uses the “High-Tech Index” which is a combination of (1) the share of the local high-tech employment (or output) regarding the total national high-tech employment and (2) the LQ. It is supposed that the index is less influenced by the effect of large or small regions. However, our final model takes a log-linear form so that logarithms have the same effect of smoothing (the correlation between both indexes is above 90%). We finally used the traditional LQ because the interpretation is easier and produce better and more robust results than the High-Tech Index.

In the estimations for Italy, the average firm size in the LPS and the average firm size in creative industries in the LPS are highly collinear, so that both have been estimated in separate regressions. Although the coefficient for the average firm size in the LPS in Italy is negative, as in Spain (although close to zero), this could be due to the collinearity of this variable with the variable “filière”.

The components of talent (human capital and creative class) are potentially endogenous. The results of a path analysis in partial regressions indeed suggest this possibility for Spain, although not for Italy. A Wu–Hausman test has been performed, testing the possible effects on the consistency of the estimations in both countries. The test confirms the results of the path analysis and suggests that creative class can be treated as exogenous in both countries, whereas the percentage of tertiary graduates could be endogenous in Spain in the partial regression (although not in the full regression). Generalized method of moments regressions have been also estimated for Spain and are available upon request. However, the effects of potential endogeneity do not hold in the full regression, and the coefficient for the percentage of tertiary graduates is unusually high in IV estimations. Consequently, in our opinion, the results of the Wu–Hausman test and the subsequent IV estimations are affected by the misspecification of the partial model, and the provided coefficient from ordinary least squares (OLS) estimations is more correct.

In this case, the effects of collinearity were demonstrably low (this was checked by including only one of the variables in separate regressions every time); so we decided to include both indicators (creative class and tertiary graduated) in the same regression.

The significance of the creative class was absorbed in the full model by other variables with which it was highly correlated (firm size, diversity, etc.).