The regional environment and firms’ commitment to innovation: empirical evidence from Spain

ABSTRACT

Using panel data for Spanish firms, this paper examines the role of firms’ regional knowledge environment for commitment to innovation of innovation active firms. The empirical results show that a stronger regional knowledge environment increases the likelihood that an innovator engages continuously in R&D as opposed to occasionally. This positive relationship with the regional knowledge environment is robust and significant in manufacturing and specifically for small firms and for technology-intensive firms.

KEYWORDS:

• Innovation
• commitment
• continuous innovation
• regions

JEL CODES:

• O31
• R1
• D22

Acknowledgements

I would like to thank Ruth Rama for helpful suggestions.

Disclosure statement

No potential conflict of interest was reported by the author(s).

ORCID

Adelheid Holl http://orcid.org/0000-0003-4045-9468

Notes

1 I would like to thank on of the anonymous referees for drawing attention to this point.

2 Other studies on innovation persistence have also included some type of regional controls (for Spain, see, for example, Máñez et al. 2009; Lopez-Garcia and Montero 2012; Triguero and Córcoles 2013) to test for regional spillovers. However, in these studies the regional variables relate to the likelihood of having introduced an innovation or having any positive amount of R&D and not specifically to persistence or the degree of commitment to innovation.

3 PITEC has different representative sub-samples: (a) companies with 200 or more employees, (b) companies with less than 200 employees which perform internal R&D, (c) companies with less than 200 employees that undertake external R&D but not internal R&D, and (d) companies with less than 200 employees and no expenditure on innovation. In 2014, the composition of the sample changed with an important drop of small firms. With the aim to reduce the data collecting burden a sample rotation has been introduced keeping some firms ‘dormant’ and not surveyed in certain years. For the year 2014, this means that about 40% of small firms that responded in 2013 were not included anymore in the 2014 survey, while nearly all medium sized firms and all large firms continued to be surveyed. In robustness check, all models have been re-estimated without the year 2014. Main results remain qualitatively the same and are available upon request.

4 Firms with extraordinary events such as mergers or acquisitions have been excluded.

5 Firms are also asked about the regional distribution of their R&D employees. Results are qualitatively the same if this information is used instead.

6 The final sample has information for 9364 firms. The number of observation for each Autonomous Community ranges from more than 14,000 in the case of Catalonia to slightly over 200 for Balearic Islands. The average number of observations per Autonomous Community and year is nearly 300, but for smaller regions, such as the islands and Cantabria, and Extremadura which suffers from a low business population, the number of observations per year is generally below 50. This should be taken into account in the interpretation of results, as signs and significance of the coefficients will be driven by those regions with greater numbers of observations.

7 The dependent variable is based on the PITEC question regarding internal R&D and takes on the value of 1 if the firm has reported carrying out R&D continuously in the survey year and 0 if the firm has reported carrying out R&D occasionally.

8 A note of caution must be raised in the interpretation of these results as reverse causality cannot be ruled out. Greater innovation engagement could lead to higher productivity or better exporting possibilities (Cassiman and Golovko 2011). Firms with greater innovation engagement are also likely to apply more for public funding. Moreover, the information on public R&D funding is not available for 2004 and for 2006, which thus reduces the estimation sample size in column 4.

9 Note, domgroup and foreign have virtually no time variation and are therefore treated as time-invariant. The variable fonpubli is not included here as the information is missing for 2004 and 2006. However, in unreported estimations with fonpubli included, main results are qualitatively the same. These results are available upon request.

10 Small firms are defined as those with fewer than 50 employees, medium-sized firms are those with 50–200 employees, and large firms are defined as those with 200 and more employees.

Additional information

Funding

I gratefully acknowledge the support of the Spanish Ministry of Education, Culture and Sports through the Salvador Madariaga Programme (PRX17/00042). Project CSO2016-79045-C2-1-R (Spanish Ministry of Economy, Industry, and Competitiveness) is also gratefully acknowledged.