What drives the choice of the type of partner in R&D cooperation? Evidence for Spanish manufactures and services

ABSTRACT

We analyse the heterogeneity in firms’ decisions to engage in R&D cooperation, taking into account the type of partner (competitors, suppliers or customers, and research institutions) and the sector to which the firm belongs (manufactures or services). We use information from the Technological Innovation Panel (PITEC) for Spanish firms and estimate multivariate probit models corrected for endogeneity which explicitly consider the interrelations between the different R&D cooperation strategies. We find that placing a higher importance to publicly available information (incoming spillovers), receiving public funding and firm size increase the probability of cooperation with all kind of partners but the role is much stronger in the case of cooperative agreements with research institutions and universities. Our results also suggest that R&D intensity and the importance attributed to the lack of qualified personnel as a factor hampering innovation are key factors influencing positively R&D cooperation activities in the service sector but not in manufactures.

KEYWORDS:

• R&D cooperation
• choice of partners
• industrial sector
• service sector
• innovative Spanish firms
• multivariate probit model

JEL CLASSIFICATION:

• O32
• L24
• C35
• L60
• L80

Acknowledgements

Erika Badillo wishes to acknowledge the financial support from the AGAUR (Generalitat de Catalunya) through ‘the grant for universities and research centres for the recruitment of new research personnel (FI-DGR 2011)’. Rosina Moreno acknowledges financial support provided by the Ministerio de Economía y Competitividad for the project entitled ‘Innovation and territory. Technological collaboration, international competitiveness and location factors’, ECO2014-59493-R, as well as for the project ‘Redes de colaboración tecnológica e innovación. Determinantes y efectos sobre la competitividad de las empresas españolas’ funded by the Fundación BBVA Ayudas a Proyectos de Investigación 2014. Thanks are also due to the Program ICREA Acadèmia.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ This database is available to the public at http://icono.fecyt.es/PITEC/Paginas/por_que.aspx.

² Note that, although PITEC has a panel structure, we carried out a cross-sectional analysis because of the complexity of the estimation strategy, as discussed in section “Method of estimation”.

³ Firms that report confidentiality issues, mergers, closures, employment incidents and so on are eliminated, as are those observations that present anomalies such as firms with zero business levels or excessively high values of R&D intensity, measured as the ratio between R&D expenditure and turnover (the rule used was the mean plus twice the SD).

⁴ That is, firms that have introduced innovations in products or processes, or who were undertaking innovation activities during the analysed period or abandoned them.

⁵ The survey also offers information on another type of cooperation: cooperation with firms in the same group. However, we do not consider such typology since only firms belonging to a group can cooperate within their group, while all the other types of partners can be chosen by all firms. However, in order to control for possible different behaviour of such firms, the regression analysis includes a dummy variable for firms belonging to a group. The same kinds of cooperation are used in López (2008) and Abramovsky et al. (2009).

⁶ Note that firms can choose not to cooperate in all cases.

⁷ We include a binary sector variable (1 = manufactures and 0 = services) in the model for the whole sample and industry dummies at two-digit level according to NACE-93 in the separate models for manufactures and services.

⁸ A more detailed explanation of several types of marginal effects can be found in Sodjinou and Henningsen (2012).

⁹ Despite lagging the explanatory variables, as pointed out by Belderbos et al. (2004), if cooperation is persistent, the factors determining this cooperation are still partly affected by those R&D agreements that were formed in the past and still in existence in the current period.

¹⁰ In addition, following Pakes’s (1983) study and some recent literature (Arvanitis and Bolli 2013), we also estimated our models using the industry averages for each of the potentially endogenous variables as the only instruments; that is, excluding export intensity and basicness of R&D variables of the set of instrumental variables. Overall, the main results remain. They can be provided by the authors upon request.

¹¹ Note that the partial R² of the first stage is very low and the value of the F-test statistics is well above 10, which is usually considered a good threshold, and so the instruments cannot be judged as weak.

¹² In order to adjust the coefficients’ SEs for the use of generated regressors, we have obtained them through bootstrapping.

¹³ As pointed out by Belderbos et al. (2004), the multivariate probit model takes these correlations into account, without being able to distinguish between the two sources of correlation. In this regard, a formal test of complementarity as in Mohnen and Röller (2005) would be necessary, but it is beyond the scope of this article.

¹⁴ See Sodjinou and Henningsen (2012) for an application in the case of the interrelations between different technology adoption decisions.

Additional information

Funding

Erika Badillo wishes to acknowledge the financial support from the AGAUR (Generalitat de Catalunya) through ‘the grant for universities and research centres for the recruitment of new research personnel (FI-DGR 2011)’. Rosina Moreno acknowledges financial support provided by the Ministerio de Economía y Competitividad for the project entitled ‘Innovation and territory. Determinantes y efectos sobre la competitividad de las empresas españolas’ funded by the Fundación BBVA Ayudas a Proyectos de Investigación 2014.