Knowledge that Transforms

We study research joint ventures (RJV) in a setting where knowledge spillovers increase with the technological proximity between firms. The scenarios we investigate differ in the intensity of collaboration which depends on the (non)coordination of research activities and the extent of exchanged knowledge. Firms can form bilateral RJVs what leads to the existence of insiders (collaborating firms) as well as outsiders (non-collaborating firms) in an oligopolistic market. Our central findings are (i) RJVs do not generally outperform competitive research with respect to innovative output and social welfare; (ii) technological proximity and the intensity of collaboration play a decisive role for the private and social benefits of a RJV; (iii) joint research combined with complete knowledge sharing does not generally outperform less intensive collaboration forms.

We are very grateful to the editor of this virtual Special Section, Ben Martin, to open up the discussion on our proposal of partly random selection of publications (Osterloh and Frey, 2020). At first glance, our proposal might look crazy. Consider, however, the quote attributed to Albert Einstein: “if at first an idea is not absurd, then there is no hope for it”. Our reply takes up three considerations discussed by the commentators Oswald (2019), Wooding (2019) and Yaqub (2019): (1) additional aspects to our paper; (2) the discussion about how bad the Journal Impact Factor (JIF) is if applied to evaluate single articles; and (3) how to overcome JIF.

This study adds to the product innovation literature by emphasizing the important yet understudied role of price distribution in shaping product demography (i.e. new product introductions and exits). While prior research has focused on market niches in the technological and geographic spaces in order to explain product demography, price space has received very limited attention despite the important role of price in the market. We posit that product dynamics are largely shaped by the existing price distribution. More specifically, we argue that local density in price space determines both the likelihood of existing products exiting the market and the rate of new products entering it. Analyzing product exit and entry in the U.S. workstation industry from 1980 to 1996, we find that while price density increases an existing product’s exit rate, new products are also more likely to enter the niches where the price density is high. We also draw attention to internal price density within multiproduct firms, analyzing a product’s price distance from the other products launched by the same firm. We find that this type of internal price density decreases both existing products’ exit rate and new products’ entry likelihood. Our emphasis on price space contributes to the literature on product innovation and demography.

The giraffe's laryngeal nerve is an absurdity. It connects the brain to the voice box, taking a circuitous route all the way down the neck and all the way back up again. Presumably, somewhere in the course of history, it got caught up with the development of other organs, and now takes a massive detour round the back of the aorta near the heart. It is anything but elegant, and was clearly “not meant to be, but just happens to be” (Gould, 1993, p76). Disentangling the nerve would involve the rearrangement of multiple organs simultaneously, and would be – to use policy parlance – a multi-stakeholder issue. The journal impact factor (JIF) is an absurdity. It connects two weak bodies of understanding – citation theory and peer review theory – via a circuitous route through the research system. The JIF was supposed to assist librarians in curating their collection, and the underlying Citation Index1 itself was supposed to help with searches and retrieval, but somewhere in the course of history, they both got caught up in an information flood. They are now used to rank selectively, a purpose for which they were certainly not designed. JIFs have ended up deeply entangled in modern research systems, and implicated in important pathologies of those systems. Of course, the nerve remains functional for most giraffes, which is more than can be said about JIFs in the research system, so the metaphor is a strained one. Even a cursory look at the JIF reveals many reasons why it does not serve us well for ranking journals. It is a metric that is unforgiving of disciplinary differences and journal styles, and easily gamed by strategic authors, editors and publishers, to name just some of the issues (Archambault and Larivière, 2009; Braun, 2012; Martin, 2016). Perhaps most troubling is the idea that journal rankings may now be shaping the direction and content of science (Rafols et al., 2012; Muller and Rijke, 2017). Authors might complain about what is needed to get published in high-JIF journals, whilst editors might lament that submissions are becoming more homogenous. Journals certainly seem to shape and influence researchers’ behaviour. Economists might not be willing to give their right arm for a publication in American Economic Review with its lofty stature, but the strength of their preferences imply they would at least be willing to sacrifice more than half a thumb (Attema et al., 2014)!2 Yet, the metaphor does serve to illustrate a choice for addressing the current malaise in research policy. Should we ‘muddle through’, à la Lindblom (1959), tinkering with modified JIFs, using them in combination with other metrics, playing the arms race with those who seek to game them?3 Or should we seek radical disentanglement from JIF, or indeed from journal rankings by any measure? Within this context, Osterloh and Frey (2020) do us an important service by asking why, for all its problems, journal rankings based on JIFs are still so influential. Moreover, they offer some suggestions for reform, some more radical than others. In this note, I wish to focus on two points of difference between Osterloh and Frey, and myself. The first is the degree to which JIF lock-in can be explained exclusively by ‘borrowed plumes’. The second is the degree to which their suggested reforms are sufficiently radical but too narrow in scope. Considering these two points, which turn out to be somewhat intertwined, will help to characterise what is really at stake. What is at stake is not so much how journal rankings affect academics and their careers (“top publications are decisive for academic careers”), but rather how well the research system interacts with and serves societal goals. Osterloh and Frey may well have identified one source of JIF lock-in, but I shall submit that there are other culprits, too, both within and beyond academia. Some are more culpable than others, readers may decide.

Should journal editors and grant-giving bodies ever make use of random-draw mechanisms to make their final decisions? This short Note argues that the answer is yes. It describes a mathematical rationale for such randomization. Put intuitively, random draw should be used when the gains from the acceptance of unorthodox path-breaking papers outweigh the costs of publishing the unorthodox poor papers. The mathematical case for this rests on an averaging argument and requires ‘convexity’ (of scientific influence) in a sense explained in the Note. The long-observed skewness of citations distributions is consistent with such convexity. Hence this Note attempts to offer a conceptual and practical addendum to the potentially important randomization proposal of Margit Osterloh and Bruno Frey (2020).