Coauthorship and Thematic Networks in AAEP Annual Meetings

We analyze the coauthorship production of the AAEP Annual Meeting since 1964. We use social network analysis for creating coauthorship networks and given that any paper must be tagged with two JEL codes, we use this information for also structuring a thematic network. Then we calculate network metrics and find main actors and clusters for coauthors and topics. We distinguish a gender gap in the sample. Thematic networks show a cluster of codes and the analysis of the cluster shows the preeminence of the tags related to trade, econometric, distribution/poverty and health and education topics.

💡 Research Summary

This paper conducts a comprehensive social‑network analysis of the Argentine Association of Economic Professors (AAEP) Annual Meeting papers from its inception in 1964 through 2023. Two complementary networks are built: a co‑authorship network, where each author is a node and an undirected weighted edge represents a shared paper; and a thematic network, where each JEL code is a node and an edge connects two codes that appear together on the same article. The authors first compile a clean dataset of titles, authors, affiliations, years, and the mandatory pair of JEL codes for every paper, standardizing author names and inferring gender from names and public profiles.

In the co‑authorship network, standard centrality measures (degree, betweenness, closeness) and community detection via the Louvain algorithm reveal a clear evolution. The early decades (1960s‑1970s) are dominated by a tight core of a few prolific scholars. From the 1980s onward, the number of active researchers expands dramatically, giving rise to multiple medium‑size clusters and a “small‑world” structure with decreasing average path length. High‑betweenness scholars act as bridges between clusters, facilitating the diffusion of ideas across sub‑fields. Gender analysis shows that women constitute only about 18 % of all authors; they have significantly lower average degree and betweenness (p < 0.05) and tend to cluster in peripheral communities, especially those focused on distribution/poverty (JEL D) and health/education (JEL I).

The thematic network exploits the fact that each paper carries exactly two JEL codes. By linking co‑occurring codes, the authors generate a weighted undirected graph whose structure mirrors the intellectual landscape of the AAEP community. Centrality analysis identifies four dominant code groups: “F” (International Trade), “C” (Econometrics), “D” (Distribution, Poverty, Income), and “I” (Health and Education). These codes form a high‑modularity cluster that persists throughout the entire period. A temporal slice reveals that while trade and econometrics remain consistently central, the health/education cluster experiences a sharp rise after the early 2000s, indicating a growing scholarly interest in welfare and human capital topics.

Methodologically, the study relies on R packages ‘igraph’, ‘tidygraph’, and ‘ggraph’ for network construction, metric computation, and visualization. Statistical validation of gender differences uses t‑tests, and the robustness of community detection is checked by comparing modularity scores across multiple runs.

Key insights include: (1) the AAEP co‑authorship structure has transitioned from a elite, tightly knit core to a diversified, multi‑clustered network, reflecting broader trends toward interdisciplinary collaboration; (2) a persistent gender gap exists both in participation rates and in network positioning, suggesting the need for targeted policies such as scholarships, mentorship, and women‑focused sessions; (3) the thematic analysis confirms the long‑standing prominence of trade and econometrics while highlighting the recent emergence of health and education as a major research frontier; (4) the JEL‑code network provides a scalable framework for mapping research topics, useful for conference organizers, funding agencies, and scholars seeking to identify emerging fields.

The authors conclude by recommending that AAEP organizers foster cross‑cluster interactions (e.g., interdisciplinary workshops), implement measures to increase female representation, and adjust call‑for‑papers strategies to balance traditional strengths with the rising demand for welfare‑related research. The paper demonstrates how longitudinal network analysis can uncover hidden patterns in academic communities and guide evidence‑based policy decisions.