The largest banks in the world have, over the past five years, signed a remarkably similar suite of climate commitments. Net-zero by 2050. Interim sectoral targets in oil and gas, power, automotive, real estate, steel, aluminium, agriculture. Glasgow Financial Alliance for Net Zero membership. Annual climate reports.
For an RI analyst building a fund’s bank engagement strategy, the question is not whether a bank has made these pledges — almost all the large names have — but whether the pledges have changed lending behaviour in ways the analyst can measure. The academic literature on this question is converging. Three papers, taken together, give you the empirical base for what is now a fairly direct verdict: not yet.
This is not “banks are lying.” It is the more interesting and useful claim that bank climate commitments have, so far, produced more disclosure than allocation change — and that the gap between stated targets and observable lending is wider than the marketing implies. If you are an RI analyst, this is the literature you cite when challenging a bank’s transition narrative. If you are a banker working inside one of these institutions, this is the literature that should inform your own institution’s next round of target-setting.
1. Sastry, Verner and Marques-Ibanez (2024) — Business as usual
The Sastry et al. paper is the cleanest empirical test in the literature of whether bank climate commitments have changed lending behaviour. The authors compare loan-level data for banks that have signed onto net-zero alliances against banks that have not. They find no measurable difference in lending to high-emission sectors after the climate commitment. Banks that committed continued to lend to oil and gas, power, and other high-emission sectors at broadly the same rate. The paper’s title — Business as usual — is the verdict.
The paper also looks at engagement. Did committed banks engage more intensively with high-emission borrowers post-pledge? The evidence is weak. Engagement frequency increased modestly; engagement depth (measured by board-level commitments from the borrower) did not.
Why this matters for your work. If you are an RI analyst engaging with a bank on its climate commitment, this is the foundational paper. The argument is no longer “banks need to set targets” — almost all of them have. The argument is “the targets have not, in observed lending data, changed behaviour.” That is a more pointed and more useful engagement frame. Ask the bank to explain the gap between their committed pathway and their realised lending.
2. ECB (2023) — Net-zero commitments by the world’s largest banks
The ECB’s occasional paper is the methodology critique. The authors examine the quality of net-zero commitments across the world’s largest banks and find wide variation in scope, baseline year, target year, sectoral coverage, and accounting methodology. Two banks with apparently identical headline pledges (“net-zero by 2050”) can mean very different things in practice: one may include capital markets activity, the other only on-balance-sheet lending; one may use a 2019 baseline, the other 2021; one may cover Scope 3 financed emissions in oil and gas, the other not.
The paper argues — and the conclusion is hard to dispute — that the absence of harmonised methodology means that bank-to-bank comparison of net-zero pledges is currently largely meaningless. The pledges are not fungible, even across banks of similar size and geography.
Why this matters for your work. This paper is the citation for the argument that bank climate commitments cannot be aggregated, ranked, or simply “checked off” without methodological detail. For an RI analyst building a comparative scorecard across banks in your fund, this is the warning that headline targets are not enough — you need to assess scope, baseline, accounting, and coverage at each bank. The paper effectively writes your due diligence checklist.
3. The Economics of Net Zero Banking (2024)
The NBER paper provides the theoretical frame for why the empirical results in Sastry et al. (2024) and ECB (2023) come out the way they do. The authors model the incentives facing a bank that has committed to net-zero financed emissions. The conclusion is that, under realistic assumptions about portfolio composition, market share, and regulatory constraint, a unilateral bank pledge produces only modest emission reductions and significant reallocation — high-emission borrowers shift to non-pledging banks rather than transitioning their underlying business. The authors call this “carbon leakage” within the banking system.
The implication is structural: net-zero banking, in the absence of either a binding regulatory framework or a coordinated industry-wide action, is largely a portfolio-rebalancing exercise. It can change which banks finance which emissions; it does not, on its own, change how much emission is financed in aggregate.
Why this matters for your work. This is the paper for the harder version of your engagement. If a bank says “we’ve reduced our financed emissions by X%,” the right follow-up question — supported by this paper — is “where did the loans go?” If they migrated to a non-pledging bank, the bank’s pledge has not, in any meaningful sense, reduced emissions. For RI analysts, this is the structural case for industry-wide engagement and policy advocacy alongside bank-by-bank stewardship. A pledge from one bank, in isolation, may not do what the marketing implies.
How to read these three together
These three papers form a tight argument:
1. Sastry, Verner & Marques-Ibanez (2024) — bank climate pledges have not, in observed lending data, changed behaviour. Empirical. 2. ECB (2023) — bank pledges are not methodologically comparable; aggregating them is misleading. Diagnostic. 3. NBER (2024) — even sincere unilateral pledges produce mostly reallocation, not absolute reduction, absent industry-wide coordination. Structural.
The three together give you an evidence-based answer to the practitioner question, “what does a bank’s climate pledge actually mean?” The answer is: in isolation and at current methodological standards, less than the marketing implies. This is not a claim of bad faith on banks’ part. It is a claim about what the institution of unilateral net-zero pledging — without harmonised methodology and without industry-wide coordination — can structurally deliver.
What you can do with this on Monday
- For RI analysts engaging banks: lead with Sastry et al. (2024). Ask the bank to explain the gap between their pledged pathway and observable lending data. The literature gives you the standing to make this challenge directly, and the bank’s response is itself a stewardship signal. - For RI analysts building bank scorecards: use the ECB (2023) framework. Score banks not on the headline pledge but on scope, baseline, sectoral coverage, and accounting methodology. Cite the ECB paper as the methodological basis. - For policy advocacy: the NBER (2024) paper is the citation for arguing that voluntary individual-bank pledges are structurally insufficient and that regulatory or industry-wide coordination is required. This is a serious policy argument, not a rhetorical one. - For sustainable finance bankers inside committed institutions: these papers are also the empirical case for your bank investing more deeply in transition finance products, sectoral pathways, and client engagement programmes that go beyond the disclosure baseline. The current model is academically vulnerable; the next iteration needs to be more substantive.
Each paper has a permanent XFID. The findings will remain relevant as the methodology debate continues to mature.
A note on the related literature
The bank-climate literature in XframeworkID is smaller than the green bond literature, but it intersects with several other strands. Two papers in the registry that are not on this list but worth reading alongside:
- The EU GBS Plausible Response paper (2023) — for how the legal-academic critique of green bonds applies, by analogy, to bank climate disclosure. - The Pricing of Green Bonds ECB paper (2022) — for how the eurozone supervisor thinks about labelled-bond pricing, which informs how it thinks about banks issuing labelled bonds.
The full corpus is at xframework.id/registry?type=RESEARCH. Each paper’s permanent XFID can be cited durably in fund prospectuses, IC papers, and policy submissions.
Part 5 of 8. Previous: Why Issuers Issue, How Issuers Disclose. Next: Sovereign and Emerging-Market Green Bonds. Browse all RESEARCH papers at xframework.id/registry?type=RESEARCH.