Out of the crisis: an empirical investigation of place-specific determinants of economic resilience

ABSTRACT

This article attempts to isolate the structural characteristics that affect the resilience of a regional economy. It focuses on the role played by related/unrelated variety and differentiated knowledge bases as drivers for regional resilience and originally explores their interdependences. Italy is the empirical setting, and Italian local labour systems the unit of analysis. Regional resilience is measured as growth of the employment rate after the Great Recession that began in 2008. Results confirm the importance of related variety and of differentiated knowledge bases as drivers for regional resilience. We found support of the creative capacity of culture argument, providing evidence that a moderate concentration in symbolic knowledge-based economic activities contributes to resilience. Synthetic and analytical knowledge-based activities provide positive and no support to regional resilience, respectively. Finally, the relatedness of the symbolic knowledge-based activities increases regional economic resilience. Some policy implications are then derived from these findings.

KEYWORDS:

• Regional resilience
• related variety
• differentiated knowledge bases
• local labour systems
• industrial districts
• economic crisis

Acknowledgments

We gratefully acknowledge valuable and stimulating comments from Philip Cooke, received during a Workshop on Resilience, cultural heritage, innovation and local development, held in Florence, the 20^th of September 2013. We are thankful to Bjorn Asheim, Ron Boschma, Fiorenza Belussi, and other participants at the AISRe Conference 2014 in Padova for their precious suggestions. We also appreciated comments from participants to the AAG Annual Meeting 2015 in Chicago. Finally, the editor and the anonymous referees provided precious comments that helped a lot improving the manuscript. The usual disclaimer applies.

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Silvia Rita Sedita http://orcid.org/0000-0002-4589-6934

Ivan De Noni http://orcid.org/0000-0001-8746-6572

Notes

1. That is, Association of American Geographers 2010, Royal Geographical Society-Institute of British Geographers 2010, European Regional Science Association 2012.

2. Nomenclature des Unités Territoriales Statistiques.

3. The concept of cognitive proximity extends in a different direction the concept of cognitive distance proposed by Noteboom (2000).

4. For a comprehensive discussion on knowledge bases and sectorial patterns of innovation, see Martin (2012c).

5. Different from the performance variable, which is measured between 2009 and 2013; all structural dimensions of regional economies are computed by using data before the crisis (2007).

6. The three knowledge bases need to be intended as abstracted ideal types because, in practice, no local economy structure is characterized by just one of the knowledge bases. Moreover, the degree to which a specific knowledge base dominates varies and depends on firms, industries and activities’ features (Asheim & Hansen, 2009). Some of the knowledge specializations overlap, influencing the empirical results. We acknowledge that the classification adopted includes some possible biases; nevertheless, after discussions with peers and colleagues at seminars and conferences, this is at the moment the most suitable method for describing the dominant knowledge base of an industry in Italy, especially if all three knowledge bases are willing to be classified using the same method.

7. Some SNA categories grouping more than one NACE class were excluded when it was unable to uniquely assign the reference NACE classes to the same prevalent knowledge base. For instance, the SNA category E (water supply; sewerage, waste management and remediation), grouping NACE classes from 36 to 39, was excluded because class 36 is inclined to be classified as relying on an analytical knowledge base, whilst classes 37, 38 and 39 are more inclined to be classified as being reliant on a synthetic knowledge base.

8. A similar analysis has been conducted by using the unemployment growth rate as dependent variable. Results, reported in Table A1 in Appendix, do not show any relevant difference. Therefore, Section 3.4 focuses on the model estimation where employment growth is the dependent variable.