Exploring and comparing innovation patterns across different knowledge intensive business services

Abstract

Using data from a survey of 769 firms, this paper provides empirical evidence of the nature of innovation and its determinants within knowledge intensive business services (KIBS). The aim of the paper is to analyse how KIBS innovate and whether they innovate differently in three Canadian knowledge intensive business industries: Computer System Designs and Related Services; Management, Scientific and Technical Consulting Services; and, Architectural, Engineering and Related Services. There are clear differences in the innovation profiles of the three sectors, which suggest that KIBS cannot be analysed as an undifferentiated group of establishments. However, there are also important within-sector differences that call for further investigation.

Keywords:

• KIBS
• Computer System Designs and Related Services
• Management, Scientific and Technical Consulting Services
• Architectural, Engineering and Related Services
• innovation
• system of innovation
• Quebec

JEL Classification :

• L84
• O3

Acknowledgements

The author acknowledges the financial support from the Social Science and Humanities Research Council in Canada (SSHRC 410-2005-0501). We would like to acknowledge the contribution of Réjean Landry and Nabil Amara to the elaboration of the survey. We would also like to thank the editor, Cristiano Antonelli, and four anonymous reviewers for their constructive comments.

Notes

According to the Oslo Manual (OECD 1997), ‘non-technological innovation covers all innovation of firms, which do not relate to the introduction of a technologically new or substantially changed good or service or to the use of a technologically new or substantially changed process’.

For the sake of concision and legibility, the abbreviated sector names, those in parentheses, will be used in the rest of the paper.

Questions were asked in French. The exact wording in French is available upon request.

A hierarchical Ward clustering procedure is used. The number of clusters is chosen so that a move to more clusters does not greatly improve the explanatory power of the classification. An eight cluster solution also corresponds to an inflection point in the semi-partial R 2.