RePEc manages quite a bit of linked metadata about publications, authors, and institutions. This allows to draw some statistics, and some of those metrics are used to compute various impact factors and rankings. In this blog post, we propose some changes in how self-citations are handled in those metrics and how the research field(s) of authors are determined. As this may have an impact on various rankings, we put those changes to a vote.
Author self-citations
Currently, any author metric using citations does not include self-citations. To be more precise, the citation count of author A does not include works authored by A (as well as citations from the same serial). This implies that a work co-authored by A and B may count differently for those two authors, unless all their works are co-authored.
The proposal is that citations by all co-authors be considered self-citations, thus the contribution to the citation count is the same for all co-authors. There is a computational benefit, the count only needs to be done once. There is also a transparency benefit, it is easier to reconcile or replicate citation numbers across authors. Finally, it (partially) deals with the problem of citation clubs.
Serial self-citations
Similarly to authors, self-citations are also not included in the computation of impact factors in the sense that citations from works in the same serial (journal, working paper, book, or chapter series) are dropped from the citation count. Many works appear in different serials, such as a journal article may also be available in a working paper series. This implies that a research work available in several serials may get a different count of citations depending on where it is listed.
The proposal is to count self-citations across all versions of a work the same way, mean that self-citations from all involved serials are dropped. The benefits are similar to those for authors: one single computation for all versions, ease of replication, and dealing with citation clubs.
Note that if both changes are adopted, the citation count for a work would be the same across all co-authors and versions.
Categorization of authors into fields
Currently, authors are categorized into fields using data from NEP. This project disseminates new economics papers, with volunteer editors choosing which ones fall into their field. Once authors have a certain number of papers disseminated through a NEP report, they are considered to be part of that field. For field-specific author rankings, the proportion of NEP-disseminated papers in that field to all NEP-disseminated papers is used to weigh their metrics.
This process has the advantage that a human editor consistently classified what is deemed to be part of a field’s literature. While not all the literature is disseminated through NEP, a good proportion is. The disadvantage is that NEP fields can change. Recently, one was split in two and new ones were added. These new reports do not have the NEP history needed to properly identify who is in that field.
The proposed solution is to include JEL codes in the classification for any work that was not disseminated through NEP. We would work with NEP editors to create a mapping between their fields and the relevant JEL codes. The advantage is that retrospective material would then be available for any new field and new material could be added for any NEP report that is dormant or abandoned.
Polls are open for a month, until 2026-04-21 00:00 UTC. Thank you for participating.
Update: Polls are closed. All three proposals have been adopted by a wide margin. They will be implemented over the coming months. Thanks to the voters for their participation.
Posted by Christian Zimmermann 
You must be logged in to post a comment.