Research Policy 33 (2004) 1355–1372 How large is the Swedish ‘academic’ sector really? A critical analysis of the use of science and technology indicators Staffan Jacobsson ∗ , Annika Rickne a Department of Industrial Dynamics, RIDE, IMIT and Chalmers University of Technology, 412 96 Gothenburg, Sweden b Department of Industrial Dynamics, Lund University, Lund, Sweden Received 11 March 2004; received in revised form 27 August 2004; accepted 8 September 2004 Abstract Sweden is perceived to be top ranking, and a ‘role model’, in terms of its volume of academic R&D. This perception is based on analyses using two standard indicators. We assess the validity of these and argue that institutional features skew the result in favour of a high ranking. Swedish academic R&D is more appropriately characterised as average, or below average, in terms of input and above average in terms of output. Science policy makers need to acknowledge this and devise policies that strengthen, rather than threaten the functioning of an efficient system. © 2004 Published by Elsevier B.V. Keywords: R&D; Bibliometrics; Indicators; Academic sector; Sweden 1. Introduction Two areas in which Keith Pavitt excelled were sci- ence policy and indicators of scientific and technolog- ical activities. He also showed a keen interest in, and a remarkable knowledge of, the small Nordic country of Sweden. In this paper, we honour him and his excep- tionally critical mind, by questioning the use of science and technology indicators to construct an image, indeed shared by himself, of Sweden as a top-ranking nation in terms of its volume of academic research (see e.g. ∗ Corresponding author. Tel.: +46 31 7721213; fax: +46 31 7721237. E-mail address: stajac@mot.chalmers.se (S. Jacobsson). Pavitt, 2001; Salter et al., 2000; S¨ orlin and T¨ ornqvist, 2000). 1 1 An earlier version of this paper was presented to the confer- ence in honour of Keith Pavitt ‘What Do We Know About Innova- tion’, held at the Freeman Centre, University of Sussex, England, November 13–15, 2003. We are grateful to Linus Dahlander, Rolf Nilsson, Olle Persson and Peter Skatt for the valuable help they gave us in preparing the first draft of this paper. We have received use- ful comments in various fora and we would especially like to thank the following colleagues for constructive comments: Mats Benner, Boel Berner, Jan Br¨ ochner, Brent Goldfarb, Charles Edquist, Anders Granberg, Staffan Laestadius and Luigi Orsenigo. We are also very grateful to three anonymous referees who provided us with extremely useful comments. Finally, the Swedish Agency for Innovation Sys- tems kindly financed part of the study. 0048-7333/$ – see front matter © 2004 Published by Elsevier B.V. doi:10.1016/j.respol.2004.09.002