Academic Research in Economics by HAKAN YILMAZKUDAY

One sentence summary: High quality collaboration is more likely to emerge as a result of an interaction between specialists and generalists with overlapping fields of expertise.

We analyze economics PhDs' collaborations in peer-reviewed journals from 1990 to 2014 and investigate such collaborations' quality in relation to each co-author's research quality, field and specialization. We find that a greater overlap between co-authors' previous research fields is significantly related to a greater publication success of co-authors' joint work and this is robust to alternative specifications. Co-authors that engage in a distant collaboration are significantly more likely to have a large research overlap, but this significance is lost when co-authors' social networks are accounted for. High quality collaboration is more likely to emerge as a result of an interaction between specialists and generalists with overlapping fields of expertise. Regarding interactions across subfields of economics (interdisciplinarity), it is more likely conducted by co-authors who already have interdisciplinary portfolios, than by co-authors who are specialized or starred in different subfields.

 

Collaboration has become the dominant mode of research production in many disciplines in recent decades. Collaboration may be motivated by career pressures to publish more and better as well as by the need to circumvent a gap of knowledge or expertise. Influence of research collaboration on citation impact is not uniform and varies largely across disciplines. However, most disciplines, including economics, reveal a strong positive correlation between citation counts and the number of co-authors. As far as economics research is concerned, co-authored papers not only have been dominating the publication scenery for several decades now but also are more likely to get accepted for publication and receive more citations than sole author papers.

In this paper, we focus on the outcome (in terms of the journal prestige and citation impact) of economists' collaborations and investigate how similarity and specialization of co-authors' research portfolios are related to the quality of collaboration. Focusing on economists provides a preferable environment for our analysis because research and collaboration in this field still takes place at a very personal level as opposed to laboratory driven research with large research teams as in many of the natural sciences. We use peer-reviewed economics journal articles between 1990 and 2014 of PhD graduates of US and Canadian economics departments whom we refer to as North American PhDs throughout this paper. This particular subset of economists can be controlled for educational background and potential social ties from the graduate school, because the American Economic Association provides full lists of all graduating North American PhDs sorted by their graduate department each year. We know that North American PhDs are a particularly influential group in academic publications: 20% of all EconLit papers, more than 50% of all papers in top general and top field journals, and about 60% of all papers in the so-called top five have at least one North American PhD on board.

Two important features in our study are co-authors' field distance and specialization levels. Co-authors with a very close field distance have publications in similar areas of economics, whereas co-authors with a large field distance have publications in different areas from one another. A concept similar to our field distance is being referred to as cognitive distance in informetrics literature. Authors' specialization levels are calculated as the Herfindahl index of their research portfolios. 

 

 

Our analysis starts with a descriptive part that yields three stylized facts on co-authors' field distance and specialization: 

Assuming a two-step process for collaborative research where co-authors search and match in the first stage and the quality of their collaboration is revealed in the second stage, we investigate the statistical significance of relations that are picked in these stylized facts. Our estimations reveal that the field distance between co-authors is negatively and significantly related to the quality of their collaborative output. This relation is robust to how quality is measured as well as whether it is the co-authors' first time collaboration or a subsequent collaboration.

There is research documenting that distant collaborations are related to better quality research compared to same location collaborations. Although the alleged importance of physical proximity between co-authors is sensitive to the nature and technological context of the research in question, and economists might still benefit from positive agglomeration effects that can be offered by large and prestigious departments with significant spillover for their colleagues in certain fields, distant collaborations have already become fairly common among economists. We find that distant and same location collaborations reveal significantly different field distances on average and since field distance is negatively related to research quality, same location collaborations are of less quality, on average. This finding complements the existing literature by providing a possible motive for engaging in distant collaboration, namely, co-authors that engage in a distant collaboration are significantly more likely to have a close field distance, and a close field distance is significantly related to having a high quality outcome for this collaboration.

Our contribution to this line of literature is to show how specialization works for and at the same time against the quality of collaboration. A high specialization level has an indirect positive effect on the quality of collaboration output because more specialized authors are more likely to team up with co-authors that have a very close field distance, and such closeness is related to a high quality of collaboration output. However, once the indirect effect is accounted for, a high specialization level has a direct negative effect on the quality of collaboration output. The total effect of specialization is negative.