EconBase tracks the econometrics literature on arXiv and links it to the wider publication record, so that a paper, its bibliography, and the people who wrote it can be looked at together.
The corpus is defined by one rule: every paper listed in arXiv econ.EM, whether econ.EM is its primary category or it was cross-listed there from stat.ME, math.ST, stat.ML, cs.LG and similar. That rule is deliberately mechanical — anyone can re-run it and get the same corpus, which would not be true of a hand-curated notion of what counts as econometrics.
For each paper we fetch the LaTeX source where arXiv has one, extract the main text from the title through the conclusion (excluding appendices), and parse the bibliography together with every in-text citation. That gives a weighted citation graph in which an edge carries not just the fact of a citation but how heavily the citing paper leans on the cited work. Roughly 8% of econ.EM submissions are PDF-only and have no LaTeX source; those can never enter this graph, and the exclusion is not random.
Author identity and publication venues come from OpenAlex, matched to our papers through a crosswalk that also merges the duplicate records OpenAlex holds for the same work (a preprint, a published article, and often several repository copies).
Each reference in a paper gets an intensity score. Mentions counts in-text citations. Sections counts how many distinct sections mention it. Main text is the share of mentions falling outside the appendix. The headline intensity is a composite of the three, on a 0–1 scale.
One caveat worth stating plainly: a majority of references are mentioned exactly once, and these all receive the same floor score. Intensity separates the handful of works a paper is genuinely built on from the many it merely nods to; it does not finely rank the latter.
EconBase is built by Martin Weidner (UCL) and Francis J. DiTraglia (Oxford), supported by UKRI grant APP47921, “Fostering a Dynamic Academic Ecosystem”. It is part of the SQARE project on research quality assessment.
The pipelines that build this site are open: see github.com/econbase/econ-corpus. Corrections and questions are welcome.