Firms’ organisational capabilities and innovation generation: the case of Italy

ABSTRACT

This paper explores the relation between firms’ organisational capabilities and their engagement in different innovation processes based on generation or adoption. We place greater emphasis on the internal generation, as this should signal whether firms’ have put their own inventive and creative efforts in the development of the new products and processes introduced and, thus, the achievement of higher capabilities of innovating. We consider that, to increase their innovative capabilities, firms could introduce different organisational innovations aimed at improving the management of their internal and external activities. Besides this, we consider whether firms are organised in a business group as also this likely shapes their organisational capabilities, enabling them to better capture and take advantage of various group-level resources for the internal generation of new technologies. The empirical analysis carried out by using recent data from the Italian Community Innovation Survey shows that different organisational innovations are always beneficial, whatever the innovation process followed by firms: however, especially those related to firms’ internal activities facilitate innovation generation more than adoption. At least to some extent, even group affiliation contributes to increase firms’ innovative capabilities, regardless of the location of the mother firm.

KEYWORDS:

• Organisational innovations
• business group affiliation
• innovation process
• generation of innovation

JEL Codes:

• O30
• O31
• O32
• O33

Acknowledgment

With the usual disclaimers, we wish to thank Claudia Capozza and three anonymous referees for their comments and suggestions.

Disclosure statement

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

Notes

1 Note: it could be not only physical or spatial proximity, but also cognitive proximity, for which it is specifically meant that firms sharing the same ‘language’, namely the knowledge base and expertise, can more easily learn from each other. Instead, if cognitive distance is too large, the actors involved might not understand each other and be able to interpret and assimilate the transferred knowledge (see Boschma 2005).

2 See appendix for details about the manufactoring sectors included in the CIS.

3 For instance, in the section on product innovation of the CIS questionnaire the second question reads: ‘Who developed these product innovations?’. The available options are: (1) ‘your enterprise by itself’; (2) ‘your enterprise together with other enterprises or institutions; (3) your enterprise by adapting or modifying goods or services originally developed by other enterprises or organisations’, (4) ‘other enterprises or organisations’. The same question is included in the subsequent section concerned with process innovation.

4 The highest correlation coefficient is that of Generation in cooperation and Generation in-house, at 0.39; the correlation of Adoption and Generation in cooperation is equal to 0.26; the lowest correlation is that between Adoption and Generation in-house, just at 0.21; all the correlation coefficients turn out statistically significant at 1% level.

5 The variable identifies firms that have indicated at least one of these first two organisational innovations identified by the CIS, that is, new business practices and/or new work methods.

6 The dataset provides information about the percentage of the firms’ employees educated (at least) to the degree level. In particular, it distinguishes between six classes as follows: 0 stands for none graduated employees; 1 for less than 5%; 2 for 5 to 9%; 3 for 10%–24%; 4 for 25%–49%; 5 for 50%–74% and 6 for 75%–100%. In the sample considered (5,442 manufacturing firms), the average class of human capital/graduated employees turns out to be 5%–9%.

7 Industry aggregations are based on the two-digit NACE (Statistical Classification of Economic Activities), see Appendix for details.

8 See Cappellari and Jenkins 2003.

9 According to Wilde (2000) the estimation of a recursive multivariate probit model, given its high non linearity, does not require exclusion restrictions for parameter identifications: indeed, there is sufficient variation in the data to identify the parameters even in case of only one varying exogenous regressor.

10 Fully consistent results are achieved by estimating a (non-recursive) multivariate probit regression, see Appendix.