How Liquid is the Green Bond Market?

Green bond liquidity

In this paper Thomas Dulak and Guntram B. Wolf reassess green bond market liquidity using a proprietary dataset from Euroclear’s LiquidityDrive, the world’s largest central securities depository which covers roughly 35% of the global green bond market as of June 2025. Unlike most prior studies that rely on dealer bid-ask quotes, this dataset captures actual secondary-market transactions: traded volumes, number of trades, trade sizes, and number of bond holders on each day.

Starting from the Bloomberg Global Aggregate Index, the authors build a matched sample of 440 green/conventional bond pairs (880 bonds in total) from 254 distinct issuers, covering the period July 2020 to June 2025. Each green bond is paired with a conventional bond from the same issuer, currency, rating, and callable status, with close proximity in issue date, maturity, and size. This matching design isolates the effect of the green label from other bond characteristics.

Liquidity is measured across five dimensions: total trading volume, whether the bond traded at all on a given day, the number of trades, average trade size, and volume at the intensive margin. Panel regressions with pair and date fixed effects are used throughout, with bond age, size, holdings volume, and number of holders as controls.

Findings by Hypothesis
H1 — Green bonds are less traded than comparable conventional bonds → Rejected (with nuance)
Green bonds are not systematically less liquid. They trade on a larger share of days, reflecting a broader, more dispersed holder base, but individual transactions tend to be smaller and the volume traded per active day is lower than for conventional bonds. The picture that emerges is not one of a thin market but of a structurally different one: frequent, small trades consistent with a buy-and-hold investor clientele that still participates actively in secondary markets.

H2 — The trading disadvantage widens during market stress → Rejected
Green bond trading does not deteriorate disproportionately when market-wide volatility rises. The green label carries no additional liquidity penalty during stress episodes, which is reassuring for investors and policymakers who worry about green bond fragility precisely when liquidity matters most.

H3 — Certified green bonds are more traded than non-certified ones → Confirmed
Third-party certification is associated with higher trading intensity as it reduces information asymmetries and greenwashing concerns, thereby attracting more active secondary-market participants. Certification does not, however, significantly increase the frequency of trading days.

H4 — Green bonds with common project profiles are more traded than niche ones → Confirmed
Green bonds financing mainstream project types, such as renewable energy, clean transport, or green buildings  trade more regularly than bonds linked to unusual or complex project profiles. This supports the intuition that a broader investor overlap around familiar project categories widens the pool of potential counterparties in secondary markets.

Greener but thinner? Assessing green bond market liquidity
Authors: Thomas Dulak, Guntram B. Wolf
From: Université Libre de Bruxelles

The sovereign greenium

Unlike the Dulak and Wolf paper above that measures liquidity through actual trading behavior (volumes, frequencies, and trade sizes) Ugo Panizza, Beatrice Weder di Mauro, Shuyang Shi, and Mitu Gulati measure liquidity through price-based costs by quantifying the bid-ask spread disadvantage of sovereign green bonds. The authors use a daily secondary-market dataset of 332 matched green/conventional bond pairs issued between 2014 and 2023 by sovereigns, sovereign-backed agencies, and multilateral development institutions. The sample covers 43 sovereign, 143 agency, and 130 multilateral bonds, with average face values of $7 billion for sovereigns and around $900 million for the other groups. For each bond the authors collect daily yields to maturity, bid-ask spreads, face value, and residual maturity.

Each green bond is paired with a conventional bond from the same issuer and currency, with identical seniority, coupon type, and rating, and with maturity and issuance date differences of no more than two years. This strict one-to-one matching isolates the pricing effect of the green label from issuer and bond characteristics.

Findings

• The greenium exists but is small. After controlling for bond characteristics, green sovereign bonds trade at a statistically significant yield discount averaging 2 basis points for advanced economies and 13 basis points for emerging markets. For sub-investment-grade issuers the greenium reaches 55 basis points, while for highly rated bonds it is positive but often not statistically significant.
• The fiscal savings are negligible. Even for a large issuer like Germany with nearly $80 billion of green bonds outstanding, a 2 basis point greenium translates into only $16 million in annual interest savings, less than 0.05% of its total interest bill.
• Liquidity disadvantage in the sovereign segment. Sovereign green bonds carry a slightly wider average bid-ask spread than their conventional counterparts (8 vs. 6 basis points), suggesting a modest secondary-market liquidity cost that partially offsets the yield advantage.
• Climate salience and vulnerability amplify the greenium. The greenium rises on days when the global transition-risk news index is elevated and is larger for countries more vulnerable to climate change.
• No binding contractual commitments. A close reading of all bond documentation finds that sovereign green bonds contain no legally enforceable obligations regarding the use of proceeds or environmental outcomes, and provide no mechanism for investors to declare default in the event of non-compliance. The greenium therefore prices symbolic rather than contractual sustainability value and its small magnitude reflects investors’ limited willingness to sacrifice returns for reputational assurances alone.

The Sovereign Greenium: Big Promise but Small Price Effect
Authors: Ugo Panizza, Beatrice Weder di Mauro, Shuyang Shi, Mitu Gulati
From: Geneva Graduate Institute, University of Virginia