While annual economics job market conferences like the AEA or AFA can be forbidding to fresh graduates anxious to secure their dream job, they also pose great opportunities to engage with colleagues about your research. Competition to present a paper at these conferences is tough and discussants are often leading experts in their ﬁeld. How important is this informal collaboration for authors to learn about new developments in their ﬁeld, or how a research paper is received by their peers?
With few notable exceptions, the existing literature studies formal collaboration – that is, co-authorship only. But co-authorship is less prevalent in economics compared to other disciplines such as biology (Laband and Tollison 2000). Consequently, when writing a research paper in economics and ﬁnance, most collaboration is informal, for example, through commentary from colleagues, feedback during seminar presentations, discussions of the paper at conferences, or even during the referee process after submitting a paper to a journal.
In a recent paper, we collect acknowledgements of 2,782 research papers published in six journals in ﬁnancial economics (Georg and Rose 2015). The Journal of Finance (JF), The Review of Financial Studies (RFS), the Journal of Financial Economics (JFE), the Journal of Financial Intermediation (JFI), the Journal of Money, Credit & Banking (JMCB), and the Journal of Banking and Finance (JBF). We look at two points in time: an early sample from 1998 to 2000, and a late sample from 2009 to 2011.
Looking at the raw ﬁgures, the intensive and the extensive margin of informal collaboration increases with the impact factor of the journal. For instance, the average JF article in 2011 acknowledged more than 12 scholars, the JBF counterpart acknowledges less than four; in 2011, every JF article acknowledges social informal collaboration, but only nine out of 10 JBF articles. The global trend between 1998 and 2011, however, is to acknowledge more informal collaboration. After manual consolidation and cleaning, we are left with 3,919 authors (of which about 50% are also acknowledged) and an additional 5,542 commenters. We connect two academics in the undirected social network of collaboration with a weight of 1 whenever they co-author a paper and additionally with a weight of 1/n whenever one acknowledges the other on a paper with n authors.1
The ﬂow of information within the profession will likely be determined by the structure of the social network of informal collaboration rather than the pure co-author network. Our network captures a dimension that Oettl (2012) terms helpfulness. Commenters spend time to review a paper, comment and make suggestions. For this reason the social network of informal collaboration contains more than twice as many researchers as a pure co-author network. Interestingly, only half of all authors are ever acknowledged, while only one out of four commenters authors a paper in our dataset. The network also connects more academics. In the late sample 98% of all economists are connected in an uninterrupted series of links in the social network of informal collaboration, while an uninterrupted path exists only for 20% of all authors in the co-author network.
Links in the general interest journals (JF, JFE, RFS) typically form the core of the network, while links in ﬁeld journals (JFI, JMCB, JBF) typically connect researchers in the periphery. Figure 1b shows the giant component of the social network of informal collaboration for the late sample. The roughly 34,000 links are colour-coded. Links from general interest journals are red, links from ﬁeld journals are blue, and the few links occurring in both types of journals are purple.
The network is heterogeneous. Some academics are much more central than others. Network centralities provide insights on the role speciﬁc individuals play in the transmission of information or the inﬂuence they exert on neighbours (Jackson 2014, Ballester et al 2006). Examining co-author networks in economics, Ductor et al (2014) write, for example: "Communication in the course of research collaboration involves the exchange of ideas. So we expect that a researcher who is collaborating with highly creative and productive people has access to more new ideas. This, in turn, suggests that a researcher who is close to more productive researchers may have early access to new ideas. As early publication is a key element in the research process, early access to new ideas can lead to greater productivity." The two most prominent centrality measures are ‘betweenness’ and eigenvector centrality. Both are related but distinct – betweenness centrality measures the importance in the transmission of information, while eigenvector centrality points to the best connected group (i.e., opinion leaders) in the network.
Figure 1. Social networks using published research articles, 2009-2011
Note: A link is drawn between every author (ﬁgure 1a, left), between an acknowledged commenter and every author (ﬁgure 1b, right) of a published research article. Red links indicate that the research article was published in a general interest journal (JF, RFS, JFE), while blue indicates a ﬁeld journal (JFI, JMCB, JBF) publication. If a link occurs in both a general interest journal and a ﬁeld journal, which is a rare event, it is coloured purple. Only the giant component is shown.
Who are the most central academics in ﬁnancial economics? Figure 2 shows the 25 most acknowledged, the 25 most betweenness-central, and the 25 most eigenvector-central academics. Higher ranked academics are in red and have a thicker line, and we show the diﬀerence in the three rankings using a rankﬂow chart.
Simply being acknowledged does not necessarily result in a central position. Out of the 25 most often acknowledged commenters, only ten are among the 25 most betweenness-central researchers, and only nine are among the 100 most eigenvector-central individuals.
Figure 2. Top 25 researchers in the social network of informal collaboration in the late sample (2009-2011)
Note: Occurrence denotes how often an academic is acknowledged, ‘Betweenness’ is the betweenness centrality, and ‘Eigenvector’ the eigenvector centrality. To compare diﬀerent rankings despite their diﬀerent ranges we use an ordinal ranking where the ﬁrst place gives 25 points, second place gives 24 points, and so on.
To understand the individual determinants of being acknowledged often or being central, we augment our dataset with individual characteristics originating from academics’ CVs. Additionally, we make use of the 2013 Tilburg Economics Department Ranking to numerically compare the quality of aﬃliations.2 Our data collection highlights remarkable diﬀerences among the typical member of each of the three sets. For example, the typical top-100 acknowledged academic is more likely to be a member of either CEPR or NBER and about three to six years more senior. The main diﬀerence between the typical top-100 betweenness-central researcher and the typical top-100 eigenvector-central researcher is that the former has been trained at a lower-ranked university and has served more years on editorial boards.3
A proportional odds model with the rank being the dependent variable reveals some diﬀerences in the predictive characteristics, at least for the three sets of top-100 academics (see Georg and Rose 2015 for detailed regression results).
- Graduating from a high ranked university is associated with being ranked high in all speciﬁcations;
- Being aﬃliated with a high ranked university has little eﬀect;
- Being female is associated with being ranked lower in all speciﬁcations;
- Being a member of NBER or CEPR is associated negatively with being acknowledged often or being central;
- Having served many years in editorial boards comes with a malus in being acknowledged often but with a statistically insigniﬁcant bonus in being central; and
- Seniority is associated with worse rankings according to the number of thanks and eigenvector centrality, but at a decreasing scale. For betweenness centrality, the opposite is true.
Authors' note: The complete ranking can be found at http://www.co-georg.de/central_places/index
Ballester, C, A Calvó-Armengol and Y Zenou (2006) “Who’s who in networks. Wanted: The Key Player”, Econometrica, 74(5): 1403–1417.
Ductor, L, M Fafchamps, S Goyal and M J V D Leij (2014) “Social networks and research output”, The Review of Economics and Statistics, 96(5): 936–948.
Georg, C-P and M E Rose (2015) “Mirror, mirror, on the wall, who is the most central of them all?”, mimeo 2709107, SSRN.
Jackson, M O (2014) “Networks in the understanding of economic behaviors”, Journal of Economic Perspectives, 28(4): 3–22.
Laband, D N and R D Tollison (2000) “Intellectual collaboration”, Journal of Political Economy, 108(3): 632–661.
Oettl, A (2012) “Reconceptualizing stars. Scientist helpfulness and peer performance”, Management Science, 58(6): 1122–1140.
1 As acknowledged academic, we deﬁne every individual that has been thanked directly for intellectual help. Speciﬁcally, we exclude industry professionals, research assistants, anonymous reviewers and/or editorial support. Moreover, we exclude known discussants and the managing editor.
3 We count the number of years an individual has been managing or associated editor.