A reader’s problem
A graduate student in 2026 wants to understand whether the green-bond market is delivering environmental impact. She begins, as her supervisor suggests, with three foundational papers: Caroline Flammer’s Corporate Green Bonds (2021), Olivier Zerbib’s Effect of Pro-Environmental Preferences on Bond Prices (2019), and Torsten Ehlers, Benoît Mojon and Frank Packer’s Green Bonds and Carbon Emissions (2020).
She finishes the Flammer paper with a question. The paper notes that issuers like Repsol are a useful test case — a Spanish energy company whose green bond issuance is, on its face, exactly the kind of anchor an empirical study can build around. But Flammer’s working sample is constructed at firm level. Where, the student wonders, would she go to read Repsol’s actual green bond framework — the document the bond was sold under? Where is the second party opinion that someone wrote on it? Has Repsol’s framework been revised since Flammer wrote? Has anyone else written about the same bond?
She googles “Repsol green bond framework PDF” and finds a thicket of fragmentary results: an SPO provider’s landing page, a 2017 issuance press release, a Climate Bonds Initiative entry, an FT article. None of it cross-references back to Flammer. None of it tells her that Julien Daubanes, Shema Mitali and Jean-Charles Rochet’s Why Do Firms Issue Green Bonds? (2022) also analyses Repsol. None of it tells her that John Patrick Hunt’s Green Bond Reporting (Columbia Business Law Review, 2023) studied the post-issuance disclosure quality of Repsol-style US-listed green bonds. Three papers, one issuer — and the only thread connecting them in the reader’s hand is whatever notes she takes herself.
This is the gap XframeworkID has just closed.
What the registry now does
Twenty-seven peer-reviewed and major-institution working papers on labelled bonds are now in the registry as a new RESEARCH document type. Each paper has its own permanent identifier (an XFID), a stored canonical PDF, and machine-readable metadata. So far, so unsurprising.
The novel thing is the connection. Across those papers, a script extracted every named bond issuer cited in the body text, cross-referenced against the registry’s 4,100-strong issuer list, and created a bidirectional link in a new document-issuer-mentions table. The result is fifty-five paper-to-issuer citations, covering thirty-eight distinct issuers across twenty papers.
What this enables is mundane until you try to do it elsewhere. On Repsol’s issuer page in the registry, there is now a section titled Research Mentioning This Issuer. It lists three papers — Flammer, Daubanes-Mitali-Rochet, and Hunt — each linked to its own permanent record, each with year and title. On each of those three paper pages, there is now a section titled Issuers Discussed. Repsol is listed, along with the other issuers each paper cites. One click in either direction.
The graduate student’s hour of googling — supposing she ever finds her way to the right three papers and the right one issuer page — is replaced by a click.
Who is studying whom
The pattern of citations, once aggregated, is a snapshot of where academic attention is currently directed.
Bank of China is the most-cited issuer in the corpus, appearing in five papers. Two of them analyse the Chinese green bond market specifically (Yang et al. 2024, Ehlers/Mojon/Packer 2020). Three are broader market studies that nonetheless flag Bank of China as a notable green issuer (Zerbib 2019, Spielberger 2024, Du/Harford/Shin 2023). For a fixed-income analyst building a view on Chinese sovereign-adjacent issuance, the registry now makes those five papers a single click from Bank of China’s framework documents.
European Investment Bank is cited in four papers — unsurprising given its 2007 EUR Climate Awareness Bond is widely treated as the ur-text of the labelled bond market. The EIB appears in Velloso and Perrotti’s ECLAC study of LATAM sustainable bonds (2023), in Spielberger’s study of EIB policy entrepreneurship around the EU GBS (2024), in Baker, Bergstresser, Serafeim and Wurgler’s foundational US municipal greenium paper (2018), and in Zerbib’s seminal greenium analysis (2019).
Repsol is cited in three papers (Hunt 2023, Daubanes-Mitali-Rochet 2022, Flammer 2021), all of which use it as an empirical anchor for studies of disclosure quality and announcement returns. Repsol’s own framework documents now sit one click from those academic critiques.
Blue Forest Conservation appears in two papers (Brand et al. 2021, Thompson 2023). Both treat Blue Forest’s Forest Resilience Bond as a case study in environmental impact bonds. The bond’s structural design report and case study, both archived in the registry, are now traversable from those papers.
Smaller mentions map onto specific research streams. Banco Sabadell, Intesa Sanpaolo, Mediobanca, Rabobank, Société Générale, Triodos and UniCredit each appear in Sastry, Verner and Marques-Ibanez’s analysis of bank climate commitments (2024). Nigeria, Hungary, Latvia, and the Republic of Korea each appear in Cheng, Ehlers, Packer and Xiao’s BIS study of sovereign green bonds (2024). Holcim Group is cited by Kölbel and Lambillon (2022) in their analysis of sustainability-linked bonds. Walmart, Kilroy Realty, and Repsol (again) are flagged by Hunt (2023) for the post-issuance reporting gap his paper documents.
The aggregate effect is a citation graph between papers and bonds that, until now, didn’t exist anywhere.
Why this matters more than it sounds
Three things follow from a citation graph of this kind, and they’re not the things that academic infrastructure usually delivers.
First, the speed-up for due diligence. A buy-side analyst at a sustainable fixed-income fund considering a Repsol bond can, in five minutes, see what every academic critique of Repsol’s green bond program has found. That is plausibly faster than the analyst’s own desk research, and it provides the analyst with a defensible, auditable trail of references for any IC paper. The registry’s permanent XFIDs mean the citations don’t break when Repsol’s IR team rebuilds their website.
Second, the speed-up for journalism. A reporter writing about labelled bond market integrity can ground a piece in academic literature without having to read every paper themselves — they can pivot from a specific bond’s documents to the academic verdict in two clicks. The names appearing in the citation graph are the names whose comment they should solicit. Caroline Flammer at Columbia, John Patrick Hunt at UC Davis, Adair Morse at UC Berkeley, Parinitha Sastry at Columbia: each is now reachable through their researcher profile in the registry, with their published work catalogued.
Third, and least obvious, the value to other researchers. A PhD student looking for an unstudied corner of the market can run the inverse query: which issuers in the registry are not cited by any paper? Which countries’ issuance has no academic literature attached? Which document types are well-studied (frameworks, prospectuses) versus thinly-studied (impact reports, EU GBS reviews)? The negative space of the citation graph is a research agenda.
What the citation graph reveals about the field
A few quiet observations, looking at the data sideways.
Real estate is under-studied at the named-issuer level. Devine, Kok, Özgür and Yönder’s REI working paper (2024) is one of the only careful empirical studies of REIT green bonds, and it constructs its sample at the building level rather than naming issuers. The result: the paper has zero issuer-level citations in our extraction. This is a weakness of academic style as much as it is a weakness of the registry’s matching logic. Researchers cite their samples as aggregates, not as named bonds. For real estate specifically, that means the paper’s findings are harder to anchor to specific issuers a journalist or investor might care about.
ISIN-level citation is essentially absent. Across all twenty-seven papers, the script found zero valid ISINs in the body text. Academic papers reference samples (“1,247 green bonds issued 2014–2022”) rather than specific instruments. The bond-by-bond connection to academic critique is therefore harder to build than the issuer-by-issuer connection. For investors who care about specific instruments, this is a gap in academic practice as much as in our infrastructure.
The geographic distribution is lopsided. Of the 38 issuers cited across the corpus, the heaviest concentrations are European banks and large multilateral issuers (EIB, IFC, IBRD). Asian green bond research focuses on China but rarely names individual non-Chinese Asian issuers. African research (Mutarindwa, Schäfer and Stephan 2024) names only sovereigns (Nigeria, Rwanda) — there is, evidently, less academic analysis of named African corporate issuers than of named European ones, despite Africa Finance Corporation, Equatorial Coca-Cola and several South African banks all being in the registry.
The biodiversity bond literature has the deepest issuer-level coverage. Thompson 2023, Brand 2021, Jeffries 2019 and Medina & Scales 2024 between them cover all five flagship nature bonds — Conservation Fund, IFC Forests Bond, TLFF, Forest Resilience Bond, Wildlife Conservation Bond — by name. This is partly because nature bonds are easier to study individually, and partly because the bonds themselves are uncommon enough to demand named treatment. The more crowded a category, the more academics resort to sample-level analysis, and the less the citation graph can connect papers to specific issuers.
What it changes
The point of all this, as far as XframeworkID’s mission is concerned, is straightforward. Labelled bond disclosure has always sat in three weakly-connected silos: the bonds themselves (and their disclosure documents), the academic literature analysing them, and the journalism and policy commentary built on top. Connecting the second silo to the first is a small change. Connecting them well — bidirectionally, with permanent identifiers, with structured machine-readable metadata — turns out to enable downstream things you couldn’t do at all when each silo was an island.
A buy-side analyst can audit her IC papers against the latest academic literature. A journalist can find the named expert whose comment best fits the angle. A graduate student can map the negative space. A regulator can find the empirical evidence that supports or contradicts a policy proposal. None of these are revolutionary actions. All of them were, until now, harder than they should have been.
There are 27 papers in the registry today. There will, in time, be more — outreach to authors of those 27 has surfaced dozens of additional candidates, and the pipeline for adding new RESEARCH-typed documents is now mechanical. The interesting question is not how many papers, but what changes in the field once academic literature on labelled bonds is permanently citable, structurally connected, and a click away from the bonds it studies.
If that question feels like a small one, the right test is to try doing the graduate student’s exercise in the existing infrastructure. The registry will look small. The web of connections it captures will not.
The full corpus of RESEARCH papers is browsable at xframework.id/registry?type=RESEARCH. Researcher profiles for the 50+ academic authors whose work is included are linked from each paper page. Readers who would like the dataset of paper-issuer citations as a CSV can request it via the contact form.
Related: Hunting the Nature Bonds · The $672 Billion 404 Error · How to Cite XFIDs