Newborn Firms and Regional Diversification Patterns: The Role of Cultural Diversity

Abstract

We blend the economics of diversity and evolutionary economic geography theories and study the relationship between the cultural diversity of foreign-born entrepreneurs and regional sectoral diversification, proxied by the sectoral variety of newborn firms. We focus on Italian evidence and use a unique data set that stems from a combination of different sources of information, including the Union of the Chambers of Commerce, the OECD, and the National Institute of Statistics. The results confirm that cultural diversity of entrepreneurs is associated with greater sectoral variety of newborn firms, with an imbalance in favor of variety in unrelated activities vis-à-vis related ones.

Key words:

• cultural diversity
• entrepreneurship
• sectoral variety
• immigrant entrepreneurs

JEL codes:

• L26
• M13
• R11
• O33

Acknowledgments

Preliminary versions of this article have been presented at the 19th Uddevalla Symposium, held at the Birkbeck Centre for Innovation Management Research, London; the annual conference of the Regional Studies Association, held at the Trinity College of Dublin in June 2017; the Geography of Innovation Conference held in Barcelona in January 2018. The authors wish to thank the participants to these conferences, as well as the editor and the anonymous referees, for their insightful comments and suggestions to improve the article.

Notes

¹ Figure 1 provides a snapshot of the relationship between the level of unrelated sectoral variety in 1999 and the average annual growth rate of gross domestic product (GDP) over the period 2000–2010, based on data about Italian NUTS 3 regions.

² The so-called Marshall-Arrow-Romer (MAR) externalities refer to the advantages stemming from the specialization of industrial activities in a given area, while Jacobs’s externalities stress the relevance of diversification as a source of location advantage for firms. In addition to these two concepts, Porter’s externalities concern the positive dynamics engendered by competitive pressures in a local context (see Antonelli, Patrucco, and Quatraro [2011] for a critical appraisal).

³ Alesina, Harnoss, and Rapoport (2016) show that the standard diversity index, which is the fractionalization index, can be decomposed into a size and a distribution component. The first is the share of the foreign-born population, and the second is a Herfindahl index computed on the shares of each country of birth over the total population. It is worth noting that size and distribution may operate in different ways. On the one hand, a larger foreign-born population increases the probability that new firms are founded. On the other hand, we expect that sectoral variety unambiguously increases if the immigrant population is more diverse in terms of countries of origin. We see at least two mechanisms through which this distributional component of diversity may affect regional diversification: one relates to the correlation between nationality and sectoral specialization and one to the possible synergies arising from the encounter of different nationalities.

⁴ The way that we defined the sectoral diversity of newborn firms, per se, applies throughout to all sectors. However, if we studied the relationship between cultural diversity and sectoral variety of newborn firms without specifying the sectors of interest, our results would likely be driven by the concentration of foreign workers in sectors such as retail trade, construction, and ethnic restaurants. More new businesses launched in these sectors, to the extent that they increase the portfolio of sectors in the region, would increase the sectoral diversification of the region according to our arguments. They would, however, add little to our understanding of the case for immigrants as possible path breakers for new development trajectories.

⁵ The flip side of this argument is that diversity based on birthplace equates second-generation immigrants with natives. Considering that second-generation immigrants are exposed to both the culture of their parents and of their country of birth, their contribution to the cultural diversity of their province of residence may be greater than the one of natives; in this sense, our measure underestimates the actual cultural diversity of entrepreneurs.

⁶ More generally, both of our measures of cultural diversity suffer from the drawback that they are based on a relatively static perspective of acculturation (Brixy, Brunow, and D’Ambrosio 2020): they equate a person’s culture with that of the country of origin, without considering that this may change during migration and that it may be the result of temporary, or circular, migration processes in different countries. Unfortunately, available data do not allow us to overcome this drawback.

⁷ While the interpretation of the immigration rate as a size component for the cultural diversity of residents seems appropriate, it may be less accurate for the cultural diversity of entrepreneurs to the extent that the entrepreneurial rates of immigrants are heterogeneous across regions. Here, the size component would be represented by the ratio of immigrant entrepreneurs over total immigrants included along with the immigration rates. We have run our full set of estimates with and without the immigrants’ entrepreneurship rates; the results are fully robust, as we show in Appendix B in the online material.

⁸ Necessity entrepreneurs are distinguished from the more standard opportunity entrepreneurs because, instead of creating businesses to grasp a business opportunity, they are forced into starting a business out of necessity due to the lack of other options in the labor market (Fairlie and Fossen 2019).

⁹ Data are available at http://demo.istat.it.

¹⁰ Detailed summary statistics can be found in Appendix A in the online material.

¹¹ This transformation is particularly useful when applied to dependent variables, since it reduces extreme values and renders the assumption of normally distributed error terms on the right-hand side reliable (MacKinnon and Magee 1990).

¹² We also report the results of two sets of regressions of the related and unrelated varieties, considered separately, in Table 5:

$NBor{n_{RV}}_{r,t}=a+{\beta }_{1}C{D}_{r,t-1}+{\beta }_{2}CD{E}_{r,t-1}+Z_{r,t-1}\gamma +{\theta }_R\omega +{\psi }_t\vartheta +{\epsilon }_{r,t}$

$NBor{n_{UV}}_{r,t}=a+{\beta }_{1}C{D}_{r,t-1}+{\beta }_{2}CD{E}_{r,t-1}+Z_{r,t-1}\gamma +{\theta }_R\omega +{\psi }_t\vartheta +{\epsilon }_{r,t}$

Although separate regressions are useful to appreciate the effects of our regressors on the different components of variety, it should be emphasized that the related and unrelated variety are inextricably linked and should not be interpreted as representing separate phenomena.

¹³ The classification is based on the standard Eurostat aggregation of NACE rev 1.1 3-digit sectors, available at https://ec.europa.eu/eurostat/cache/metadata/en/htec_esms.htm.

¹⁴ Because the coefficients of patent intensity on both components of variety are of approximately the same magnitude, the one on their ratio $NBor{n}_{RV/U{V}_{r,t}}$ is much smaller and insignificant. This implies that patent intensity correlates with both components of sectoral variety and not more strongly with either of the two.