Assessing the returns to collaborative research: Firm-level evidence from Italy

Abstract

We use firm-level data from Italian manufacturing firms to assess the relationship between various types of R&D and total factor productivity growth, including collaborative research with other firms and universities. A novel twist to our empirical analysis is that we estimate a sample selection model, which allows us to treat the decision to conduct R&D as endogenous. We find strong evidence of positive returns to collaborative research with other companies, whereas collaborative research with universities does not appear to enhance productivity. This result implies that firms may conduct R&D with universities when appropriability conditions are weak and the outcomes of such research projects do not yield direct strategic benefits.

Keywords:

• R&D
• Collaborative research
• Total factor productivity
• Sample selection bias

Notes

¹See Link and Siegel (2003) for a review of the old and new growth literatures relating to investment in technology.

²See Griliches (1990).

³See Siegel et al. (2003) for further discussion of this issue.

⁴Data from the first survey were used in Piga (2002) to study the strategic use of debt in vertical relationships, whereas the decision to conduct cooperative R&D and its antecedent decision to engage in R&D are jointly examined in Piga and Vivarelli (2004) using the 1998 survey.

⁵Piga and Vivarelli (2004) present evidence suggesting a complementary relationship between internal and external R&D activities.

⁶It should be noted that collaboration with external institutions is interpreted here in the sense that the research facilities of the external partners were used. It is not possible to identify whether the firm is outsourcing its research or whether it is joining forces with its external partners in a way that entails pooling of competences and specific know-how.

⁷The North-West region contains cities such as Turin, where the FIAT headquarters are located, and Milan, which is traditionally considered the financial capital of Italy.

⁸The geographical compositions in the two samples are similar to the one reported in Evangelista et al. (2002) who use the Italian data collected for the European project known as the ‘Community Innovation Survey’ comprising 22,787 firms.

⁹The empirical results are more heterogeneous in studies based on small samples and seem to depend on the econometric methodology adopted (cross section, panel data, etc). See Mairesse and Sassenou (1991) for additional details.